Reprint of: The Diagnostic and Prognostic Utility of Incorporating DAXX, ATRX, and Alternative Lengthening of Telomeres (ALT) to the Evaluation of Pancreatic Neuroendocrine Tumors (PanNETs)

Summary

Pancreatic neuroendocrine tumors (PanNETs) are a heterogeneous group of neoplasms with increasing incidence and an ill-defined pathobiology. Although many PanNETs are indolent and remain stable for years, a subset may behave aggressively and metastasize widely. Thus, the increasing and frequent detection of PanNETs presents a treatment dilemma. Current prognostic systems are susceptible to interpretation errors, sampling issues, and do not accurately reflect the clinical behavior of these neoplasms. Hence, additional biomarkers are needed to improve the prognostic stratification of patients diagnosed with a PanNET. Recent studies have identified alterations in death domain-associated protein 6 (DAXX) and alpha-thalassemia/mental retardation X-linked (ATRX), as well as alternative lengthening of telomeres (ALT), as promising prognostic biomarkers. This review summarizes the identification, clinical utility, and specific nuances in testing for DAXX/ATRX by immunohistochemistry and ALT by telomere-specific fluorescence in situ hybridization in PanNETs. Furthermore, a discussion on diagnostic indications for DAXX, ATRX, and ALT status is provided to include the distinction between PanNETs and pancreatic neuroendocrine carcinomas (PanNECs), and determining pancreatic origin for metastatic neuroendocrine tumors in the setting of an unknown primary.

1. Introduction

Pancreatic neuroendocrine tumors (PanNETs) are a heterogeneous group of neoplasms with variable clinical behavior. The first report of a PanNET was by Albert Nicholls, who described an adenoma arising from the islets of Langerhans in 1902 [1]. PanNETs have since been recognized as an endocrine malignancy and are the 2nd most common neoplasm of the pancreas. Historically, the incidence of PanNETs has been low and patients often present with clinical signs and symptoms of peptide hormone hypersecretion (functional PanNETs), such as insulin, glucagon, and gastrin [2]. Furthermore, these historical reports have shown most patients present with locally advanced or distant metastatic disease at the time of diagnosis. Considering patient outcome is closely associated with metastatic progression, the median overall survival for patients with PanNETs was poor with a reported 5-year overall survival rate of 32.8% [3].

In recent years, the incidence of PanNETs has steadily increased and, currently, PanNETs affect 1 per 100,000 individuals annually [4–7]. This increase in incidence is mainly represented by a higher proportion of patients diagnosed with early-stage disease (localized) and an absence of hormonal hypersecretion syndrome (nonfunctional PanNETs). It is hypothesized that the greater use of cross-sectional imaging for unrelated clinical conditions is responsible for these “incidentally” detected PanNETs. Moreover, a recent SEER registry study estimated the 5-year overall survival rate of PanNET patients to be 59.1% [7]. Considering that many nonfunctional PanNETs may follow an indolent clinical course, the overtreatment of PanNETs has been a subject of debate, and, therefore, an observational approach may be necessary for a subset of patients [8–10]. Hence, the identification of prognostic, and even diagnostic, PanNET biomarkers has been an area of intense investigation. Promising biomarkers that have received particular interest and have multiple potential clinical applications are the assessment of DAXX, ATRX, and alternative lengthening of telomeres (ALT), which is subject of this review.

2. Issues with current biomarkers for PanNET patients

Currently used clinical prognostic biomarkers for patients with PanNETs include the specific type of peptide hormone secreted for functional PanNETs, tumor size, and WHO histologic grade [11–13]. Although surgical intervention is often required for any patient with a functional PanNET to address associated clinical signs and symptoms, the risk of metastasis and overall survival is reported to correlate with the subtype of the functional PanNET. For instance, hypersecretion of insulin or insulinomas, which are the most prevalent functional PanNET, typically follow an indolent clinical course. In fact, the 20-year disease-specific survival rate of insulinoma patients approaches 93% [14]. Consequently, enucleation of an insulinoma without regional lymphadenectomy is frequently the preferred surgical technique [15–17]. The remaining functional PanNETs have a notably low incidence, but the risk of metastasis for gastrinomas, glucagonomas, VIPomas, and somatostatinomas is statistically higher and ranges from 50 to 92% [18]. Because of the rarity of noninsulinoma functional PanNETs, a large series examining disease-specific survival based on individual subtype is lacking, but gastrinomas are associated with a 5-year disease-specific survival rate of 35% [19]. Therefore, a nodal dissection of noninsulinoma functional PanNETs at the time of surgery is commonly indicated. However, approximately 10% of insulinomas can develop metastatic disease, and, thus, other factors are likely to influence the metastatic spread of functional PanNETs.

Examining the association between “incidentally” detected PanNETs and an improved 5-year overall survival rate over the last decade, tumor size has also emerged as an important prognostic factor. Recommendations by the National Comprehensive Cancer Network, European Neuroendocrine Tumor Society, and North American Neuroendocrine Tumor Society are surgical resection for large (>2.0 cm) PanNETs with regional lymphadenectomy because of their association with distant metastases, whereas a limited surgical approach or surveillance can be considered in small (≤2.0 cm) PanNETs as these neoplasms often do not metastasize [11–13]. Within a multi-institutional retrospective case series of small PanNETs, Gaujoux et al. followed 46 patients by serial imaging for a median of 41 months, and none of the patients developed lymph node or distant metastases [20]. Similarly, Sadot et al. published a matched case-control study of small PanNETs with 104 patients who were observed and compared to 77 patients who underwent upfront surgical resection [9]. Within a median follow-up of 44 months, none of the patients within the observation group developed metastatic disease and none of the patients died of disease. However, the authors did not find a difference in overall survival between patient groups. In contrast, studies by Haynes et al., Sallinen et al., and Lombardi et al. reported distant metastatic rates of 7.7%, 16.7%, and 17.4% for patients with small PanNETs [21–23]. Based on an analysis of the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database, the distant metastatic rate of small PanNETs was 9.1% [7]. Upon reviewing 1014 patients with small PanNETs from the National Cancer Database (NCDB), Assi et al. showed a higher 5-year overall survival rate for patients underwent surgery as compared to patients who were followed by surveillance [24]. For patients with 1.0–2.0 cm PanNETs, the 5-year overall survival rate for the surgical group was 92%, whereas the observational group had a 5-year overall survival rate of 70%. Of note, the authors found that surgical resection did not improve overall survival among patients with PanNETs of <1.0 cm. An optimal size cut-off for PanNETs to stratify patients at risk for distant metastasis and improved overall survival has yet to be determined.

The proliferative activity of a PanNET has repeatedly been shown to provide prognostic information. Based on mitotic index and Ki-67 immunohistochemistry, the World Health Organization (WHO) subdivides PanNETs into 3 histologic categories: low-grade (grade 1, G1), intermediate-grade (grade 2, G2), and high-grade (grade 3, G3) [25,26]. A G1 PanNET is defined as having <2 mitotic figures per 2 mm² and a Ki-67 index of <3%, a G2 PanNET has 2 to 20 mitotic figures per 2 mm² or a Ki-67 index of 3–20%, and a G3 PanNET has >20 mitotic figures per 2 mm² or a Ki-67 index of >20%. Several studies have shown that advanced histologic grades, specifically G2 and G3 PanNETs, correlate with an increased likelihood of distant metastases. Among 140 PanNET patients, Hamilton et al. identified 5-year disease-free survival rates of 95%, 80%, and 52% for G1, G2, and G3 tumors, respectively [27]. As reported by Ellison et al., 5-year overall survival rates also correlated with WHO grades: 85% for G1, 78% for G2, and 9% for G3 [28]. Median survival times were 182, 108, and 20 months for patients with G1, G2, and G3 PanNETs, respectively. However, the WHO classification scheme is largely based on the consensus of experts. Hence, it is not surprising that different cut-offs, such as a Ki-67 of 5% to differentiate between G1 and G2 PanNETs, have been proposed to improve the prognostication of patients [29]. Nevertheless, several factors can influence the evaluation of Ki-67 immunolabeling, such as the use of different antibody clones, the thickness of tissue sections, apoptotic nuclei, and infiltrating lymphocytes. Moreover, immunohistochemistry for Ki-67 can exhibit significant variability in intensity of staining, and specific cut-offs with regard to staining intensity have not been established. A final consideration regarding Ki-67 immunolabeling is the method used for evaluation: “eye-ball” estimation, camera-captured “hot spot” assessment, or digital image analysis [30–32].

Although functional subtype, tumor size, and WHO histologic grade have shown to be clinically useful in prognosticating patients and stratifying those that require surgical intervention, there are also obvious limitations with these biomarkers. The central issue is that current biomarkers represent indirect measurements of clinical behavior, and they do not accurately reflect the underlying biology of these neoplasms. As a result, many investigators have focused their research efforts on the identification of additional biomarkers that incorporate a personalized and objective approach to the prognostic evaluation of PanNETs.

3. The discovery of DAXX/ATRX mutations and the presence of ALT in PanNETs

In 2011, a landmark study characterized the genetic landscape of well-differentiated PanNETs [33]. Using exomic sequencing in an initial discovery cohort of 10 patients, all of whom but one developed liver metastases, 7 primary tumors and 3 liver metastases were sequenced. Commonly identified mutated genes were subsequently sequenced in a validation cohort composed of an additional 58 PanNETs. As predicted based on previous studies, MEN1, which codes for the tumor suppressor menin, was found to be the most frequently mutated gene (44%). Intriguingly, for the first time in any cancer type, recurrent inactivating mutations were identified in DAXX (located on 6p21.3) and ATRX (located on Xq21.1), genes which encode a chromatin remodeling complex (Fig. 1). Overall, a total of 43% cases displayed either a DAXX (25%) or ATRX (18%) mutation. Importantly, these somatic mutations were mutually exclusive, thereby implicating the DAXX/ATRX complex function as integral to PanNET pathogenesis. Subsequent whole-genome sequencing of 102 primary PanNETs revealed similar mutational frequencies and confirmed the mutual exclusivity of the gene mutations [34].

Fig. 1.

Surgically resected ATRX-mutant/ALT-positive well-differentiated pancreatic neuroendocrine tumor (PanNET). A, A classic pancreaticoduodenectomy (Whipple resection) was performed for a 2.1 cm well-circumscribed PanNET that was characterized by (B) an organoid appearance of neoplastic cells with granular cytoplasm and stippled nuclear chromatin. In this case, mitotic figures were identified at a rate of 1 mitosis per 2 mm² and (C) immunolabeling for Ki-67 showed a low proliferation index of 2%. Based on both mitotic figures and Ki-67 labeling, the PanNET was histologically classified as WHO grade 1. Despite these histologic and immunohistochemical findings, lymphovascular and perineural invasion, and lymph node metastases were seen. Additional immunohistochemical studies demonstrated (D) preserved expression for DAXX, but (E) loss of ATRX nuclear expression. Furthermore, telomere-specific fluorescence in situ hybridization detected the (F) presence of ultrabright, intranuclear foci (white arrowhead), consistent with alternative lengthening of telomeres (ALT).

Immediately following the publication of the original PanNET genetic landscape, Heaphy et al. uncovered that DAXX and ATRX inactivation, either by gene mutation of loss of protein expression, correlated perfectly with activation of ALT, a telomerase-independent and cancer-specific telomere maintenance mechanism [35]. As a histone chaperone complex, DAXX/ATRX recognizes H3K9me3 in chromatin and deposits the histone variant H3.3 at repetitive sequences. This chromatin remodeling function of DAXX/ATRX is important for heterochromatin maintenance at repetitive sequence elements, particularly at telomeric and pericentromeric regions [36,37]. These repetitive sequences are inherently unstable and susceptible to replication errors, thereby susceptible to undergo aberrant recombination. Thus, owing to telomere deprotection caused by telomere-specific chromatin alterations, ALT-positive PanNETs maintain telomere lengths through a homology-directed DNA repair mechanism. This aberrant process is similar to break-induced repair and can result in rampant chromosomal instability (including extensive genome rearrangements), defects in the G2/M checkpoint, marked micronucleation, altered double-strand break repair, and increased replication stress with resultant fork stalling [38,39]. Notably, ALT has been reported to be present in ~5–10% of all cancers, and although ALT can infrequently arise in carcinomas, ALT is particularly enriched in tumors with a mesenchymal or neuroepithelial origin (eg, sarcomas, neuroblastomas, pediatric glioblastoma) [40–42].

ALT-positive cancers feature characteristics expected from aberrant homologous recombination at telomeres, including dramatic cell-to-cell telomere length heterogeneity and the presence of single-stranded species of extrachromosomal circles containing the C-rich telomere repeat sequence (C-circles). As such, although areas exhibiting necrosis were excluded from consideration, the original criteria for ALT positivity in PanNETs included the following: (1) the presence of ultrabright, intranuclear foci of telomeric DNA FISH signals (with integrated total signal intensities for individual foci being >10-fold that of the per cell mean integrated signal intensities for all telomeric signals in individual benign stromal cells within the same case), and (2) ≥1% of neoplastic cells displaying ALT-associated telomeric DNA foci [35]. In contrast, tumors lacking ALT-associated telomeric foci in which at least 5000 cells were considered ALT-negative. Since then, these criteria have been applied by most groups, including the 1% threshold, although the number of cells assessed has varied depending on the specimen type.

4. DAXX/ATRX and ALT as prognostic biomarkers for PanNETs

With the identification of recurrent genomic alterations in DAXX and ATRX, and by association with the presence of ALT, Jiao et al. showed mutations in DAXX/ATRX correlated with a trend for prolonged overall survival [33]. Based on 66 patients, the authors found the 5-year and 10-year overall survival for DAXX/ATRX mutations to be 96% and 80%, as compared to 72% and 54% for PanNET patients who lacked these genomic alterations. The authors, however, did note their patient cohort size was small and required further validation on a larger series. In contrast, Marinoni et al. demonstrated loss of DAXX/ATRX nuclear expression, a surrogate marker for DAXX/ATRX mutations, in PanNETs was associated with distant metastasis, shorter disease-free survival, and shorter disease-specific survival [43]. The discordance between the studies by Jiao et al. and Marinoni et al. may be attributed to differences in the patient populations investigated. All patients evaluated by Jiao et al. had metastatic disease, as opposed to 18% of patients reported by Marinoni et al. But once again, the number of patients with adequate follow-up within the study by Marinoni et al. was small. In addition, correlative ALT analysis was performed for only a subset of PanNET patients.

To resolve the discordance and address the issues raised by Jiao et al. and Marinoni et al., 2 large cohort studies comprehensively evaluated the status of DAXX/ATRX and ALT among patients with a PanNET. Within a study of 269 patients, Kim et al. reported a strong association of DAXX/ATRX loss and the presence of ALT with large tumor size, the presence of lymph node metastasis, lymphovascular and perineural invasion, and distant metastasis [44]. In parallel, Singhi et al. analyzed 321 PanNET patients from 2 institutions and demonstrated DAXX/ATRX loss and ALT-positivity to be associated with larger mean tumor size, higher WHO grade (G2 and G3), lymph node metastasis, and distant metastasis [45]. The 5-year disease-free survival and 10-year disease-specific survival of patients with DAXX/ATRX-negative and ALT-positive PanNETs were 50% and 40%, respectively, as compared with 89% and 96%, respectively, for wild-type PanNETs. Among distant metastases, DAXX/ATRX loss and the presence of ALT were detected in 52% and 67% of PanNETs, respectively, and only occurred in the setting of a DAXX/ATRX-negative primary PanNET. Of note, the authors also identified a subset of PanNETs with heterogeneous loss of DAXX/ATRX and corresponding heterogeneous presence of ALT. Moreover, Singhi et al. found that PanNET patients with heterogeneous loss and heterogeneous presence of DAXX/ATRX and ALT, respectively, developed distant metastases with homogeneous loss and homogeneous presence of DAXX/ATRX and ALT, respectively. Finally, by multivariate analysis, both DAXX/ATRX loss and ALT-positivity were independent prognostic factors for poor disease-free survival.

A defining prognostic study of DAXX/ATRX and ALT status in PanNETs was conducted by Hackeng et al. and included a large multi-institutional and international cohort of 561 patients without synchronous distant metastases [46]. Similar to prior reports, the authors found loss of DAXX/ATRX and the presence of ALT correlated with several adverse pathologic features, distant metastases/recurrence on follow-up, and were independent, negative prognostic biomarkers for PanNETs. Importantly, a shorter disease-free survival was identified for patients with small (≤2.0 cm) PanNETs and DAXX/ATRX loss and/or ALT-positivity. Pea et al. also confirmed the prognostic power of ALT in a series of 87 small PanNETs, but using a 3.0 cm tumor size cutoff.

However, the aforementioned studies were restricted to surgical resection material, while the true clinical utility of DAXX/ATRX immunohistochemistry and telomere-specific FISH for ALT is stratifying patients for surgical management in the preoperative setting (Fig. 2). Several studies have explored the assessment of DAXX, ATRX, and ALT using preoperative specimens. Among 21 PanNET patients with paired preoperative and postoperative specimens, VandenBussche et al. demonstrated 100% concordance in DAXX/ATRX and ALT status [47]. Although published in abstract form, Singh et al. also demonstrated a high concordance rate between endoscopic ultrasound (EUS)-fine-needle aspiration (FNA) specimens and surgical resection specimens within a cohort of 102 PanNET patients [48]. Moreover, loss of DAXX/ATRX and the presence of ALT using preoperative specimens were associated with increased mean tumor size by EUS, higher WHO grade, distant metastases, and both decreased disease-free and disease-specific survival. Once again, the detection of DAXX/ATRX loss and ALT-positivity in EUS-FNA specimens were negative, independent prognostic factors for disease-specific survival. These confirmatory preoperative studies set the stage for incorporating DAXX/ATRX immunohistochemistry and telomere-specific FISH for ALT as prognostic biomarkers in the clinical evaluation of PanNET patients.

Fig. 2.

Fine-needle aspiration of a DAXX-mutant/ALT-positive well-differentiated PanNET. A, Cell block specimen of a PanNET that consists of loosely cohesive neoplastic cells with plasmacytoid appearance, eosinophilic granular cytoplasm, and finely stippled (salt-and-pepper) chromatin. Immunohistochemical stains demonstrate (B) loss of DAXX expression with preserved expression in non-neoplastic cells (yellow arrowhead), while preserved expression for ATRX was observed (not shown). Consistent with DAXX/ATRX loss, (C) telomere-specific FISH identified the presence of ultrabright, intranuclear foci (white arrowhead), which is indicative of ALT positivity.

5. Metastatic DAXX/ATRX-negative and ALT-positive PanNETs to the liver

As outlined earlier, ALT has been extensively validated as a poor prognostic biomarker in primary PanNETs. However, there was initial confusion on the potential clinical utility of ALT (and the presence of inactivating DAXX and ATRX mutations). In the original Jiao et al. study, the discovery and validation cohorts were enriched for advanced cases [33]. For example, of the 68 cases analyzed, the tumor size was large (4.85 cm; 2.3–24.0 cm), 41% developed liver metastases, and 66% were classified as either locally advanced or metastatic. Among patients with metastatic PanNETs, 15 patients with a DAXX/ATRX mutant PanNET (and presumably ALT-positive) had a better overall survival compared to 12 patients with a DAXX/ATRX wild-type PanNET. Likewise, Dogeas et al. showed in a cohort of 43 patients with a PanNET liver metastases, ALT positivity was a better prognostic marker in this clinical setting [49]. Thus, these findings seem to contradict latter studies showing ALT as a poor prognostic marker.

However, a study by Kim et al. may have helped elucidate the seemingly contradictory findings [44]. In this cohort of 269 patients, primary ALT-positive PanNETs were independently associated with aggressive clinicopathologic features and reduced recurrence-free survival. In contrast, ALT was associated with better overall survival in patients with metastatic lesions. Specifically, synchronous or metachronous metastatic PanNETs were observed, and within this subgroup, the patients with ALT-positive primary tumors had significantly better overall survival than those with ALT-negative tumors. Likewise, when limiting to only the metachronous metastatic PanNETs, the overall survival in patients with ALT-positive tumors remained significantly longer than in those with ALT-negative tumors. These observations strongly suggest that context matters (eg, primary versus metastases and recurrence-free survival versus overall survival) in which clinical setting ALT is used for prognostication. Future studies will need to elucidate the underlying biology of ALT-positive metastases, but these findings shed light on how to best use the assessment of ALT.

6. The utility of evaluating DAXX, ATRX, and ALT status in determining site of origin for a metastatic neuroendocrine tumor

In addition to PanNETs, DAXX/ATRX mutations and the presence of ALT have been described in a diverse number of neoplasms: glioblastomas, thyroid neoplasms, breast cancers, sarcomas, neuroblastomas, malignant melanoma, paragangliomas, hepatocellular carcinomas, renal neoplasms, prostatic adenocarcinoma, ovarian clear cell carcinoma, endometrial serous carcinomas, and urothelial carcinomas [35,40,50–55]. But, in contrast to PanNETs, the prevalence of DAXX mutations is significantly lower than ATRX alterations in other neoplasms. Among neuroendocrine tumors (NETs) of other sites, ALT-positivity has been reported in gastrointestinal tract NETs and carcinoid tumors of the lung. Heaphy et al. were the first to publish the presence of ALT in 2 of 32 (6%) gastrointestinal tract NETs. In a follow-up manuscript by Dogeas et al., the authors identified ALT in 2 of 42 (4%) metastatic gastrointestinal tract NETs to the liver [49]. For Heaphy et al. and Dogeas et al., the site of origin of the gastrointestinal tract NETs was not stated; however, the authors reported that 24 of 43 (56%) metastatic PanNETs harbored ALT. Considering metastatic NETs to the liver are likely of either gastrointestinal tract or pancreatic origin, Dogeas et al. concluded that the identification of ALT-positivity within a metastatic NET was a highly specific biomarker for pancreatic origin.

Although the intent of the multi-institutional and international study of DAXX, ATRX, and ALT by Hackeng et al. was to determine the prognostic role of these biomarkers among primary PanNETs, a secondary aim was to comprehensively evaluate NETs from other organ sites [46]. The authors’ study, therefore, included 341 primary, nonpancreatic NETs, and 313 nonpancreatic NET metastases. Within this large and diverse cohort of NETs, loss of ATRX and the presence of ALT were identified in only 2 of 48 (4%) nonmetastatic carcinoids of the lung, but not in NETs from other sites. Moreover, loss of DAXX was not observed in any primary, nonpancreatic NET. Likewise, none of the metastatic nonpancreatic NETs demonstrated DAXX/ATRX loss nor ALT-positivity. In comparison, the authors evaluated 107 PanNET metastases and found loss of DAXX/ATRX loss and the presence of ALT for 76 (71%) and 66 (62%) PanNET metastases, respectively, thereby strongly enriched in the metastases compared to primary PanNETs. Based on these findings, the status of DAXX, ATRX, and ALT may not only serve as prognostic biomarkers but also diagnostic biomarkers of pancreatic origin for a NET metastasis when a primary site cannot be clinically determined (Fig. 3). Moreover, the authors’ results indicate that DAXX, ATRX, and ALT play important roles in the molecular pathogenesis of PanNETs, but not for NETs from other organ sites.

Fig. 3.

Core liver needle biopsy of a metastatic well-differentiated neuroendocrine tumor. A, The neuroendocrine tumor was characterized by a nested growth pattern of monotonous neoplastic cells that were well demarcated from the surrounding hepatic parenchyma. Immunohistochemical stains demonstrate (B) loss of DAXX expression and (C) the presence of ALT (white arrowhead) by telomere-specific FISH, both of which are consistent with a pancreatic primary.

7. Distinguishing between PanNETs and PanNECs using DAXX, ATRX, and ALT

NETs of the pancreas not only include PanNETs but also pancreatic neuroendocrine carcinomas (PanNEC) and mixed neuroendocrine–non-neuroendocrine neoplasms (MiNENs) of the pancreas [25]. Both PanNECs and MiNENs are extremely rare pancreatic neoplasms and, in contrast to PanNETs, patients do not clinically present with symptoms because of excess hormone secretion [26]. Moreover, rather than elevated neuroendocrine markers, patient plasma studies demonstrate elevations in common carcinoma markers (eg, CA19–9 and CEA) [56]. Typical radiographic imaging modalities for PanNETs, such as octreotide scintigraphy and positron emission tomography using a tracer that is gallium (Ga) 68 labeled to the macrocyclic chelating dodecanetetraacetic acid (DOTA) are negative [57]. These neoplasms are aggressive, and most patients die within a year of diagnosis.

In most cases, the pathologic diagnosis of PanNETs, PanNEC, and MiNENs is relatively straightforward and, as per the 2019 WHO classification system, based primarily on morphologic features. However, distinguishing between G3 PanNETs and PanNECs can be diagnostically challenging. In particular, this distinction can be problematic in the absence of pertinent clinical information (eg, patient symptoms, serum studies, octreotide scintigraphy) and limited tumor specimen (eg, biopsies or cytology specimens). In a study by Tang et al. to determine the agreement rate among subspecialized pathologists to distinguish between G3 PanNETs and PanNECs, the authors observed a lack of consensus among two-thirds of cases using morphology alone, especially when only a biopsy specimen was available for review [58].

Analogous to PanNETs, whole exome sequencing and whole genome sequencing studies have identified recurrent genomic alterations in PanNECs, but rather than involving DAXX and ATRX, include mutations in TP53, RB1, and, less commonly, KRAS and SMAD4 (Fig. 4) [59,60]. In addition, TP53 and RB1 alterations often result in aberrant nuclear expression of their respective proteins. TP53 missense mutations result in a strong nuclear labeling for p53 in >80% of neoplastic nuclei, while a TP53 deletion or truncating mutation is associated with complete absence of p53 immunolabeling. Inactivation of RB1, either by homozygous deletion or intragenic mutation with loss of heterozygosity, correlates with loss of nuclear expression for Rb. Although mutations in TP53 and RB1 have been described in a small subset of PanNETs, aberrant expression for p53 and Rb, respectively, is a rare finding [61]. Thus, considering the high specificity of DAXX and ATRX for PanNETs, and p53 and Rb for PanNECs, an immunohistochemical panel of these four proteins can be useful in distinguishing between high-grade pancreatic neuroendocrine neoplasms. In fact, based on the aforementioned study by Tang et al., the authors found that in the setting where a G3 PanNET or PanNEC cannot be ascertained by clinical findings and morphology alone, 92% of cases were able to be classified using the four-protein immunohistochemical approach [58].

Fig. 4.

Representative examples of (A to C) a well-differentiated PanNET, WHO grade 3, and (D to F) a pancreatic neuroendocrine carcinoma (PanNEC), large cell type. The distinction between WHO grade 3 PanNETs and PanNECs can be challenging based on (A and D) histologic features and mitotic/Ki-67 proliferation index (B and E, respectively). In the appropriate clinicopathologic setting, immunohistochemical studies have found that loss of DAXX and/or (C) ATRX suggests a diagnosis of a PanNET, while preserved expression for DAXX and (F) ATRX is supportive of a PanNEC diagnosis.

8. The challenges of DAXX, ATRX, and ALT testing

Despite the utility of determining the status of DAXX, ATRX, and ALT in pancreatic neuroendocrine neoplasms, there are several issues that should be recognized when incorporating their assessment into clinical practice. For instance, next-generation sequencing of DAXX and ATRX is not widely available and, in many scenarios, not reimbursable for diagnostic and prognostic purposes. Moreover, loss of DAXX/ATRX does not always coincide with the presence of an alteration within the respective gene. These findings suggest additional genomic mechanisms other than mutational inactivation, such as epigenetic silencing can be responsible for DAXX/ATRX loss of protein expression. Issues may also arise with immunohistochemical labeling of DAXX/ATRX. Hechtman et al. showed a perfect concordance rate between mutations in exons 1 through 7 of DAXX and loss of DAXX expression, but a small subset of mutations in exon 8, the last exon of DAXX, often show retained protein expression [62]. Considering nonsense-mediated mRNA decay (NMD) is often inefficient due to mutations downstream from the last exon junction complex, the authors surmised that escape from NMD explains the inability of DAXX immunolabeling to be used as a surrogate marker for DAXX mutations in exon 8.

Currently, the preferred method for detecting ALT is telomere-specific FISH. This assay involves in situ hybridization of a fluorescent-labeled peptide nucleic acid probe consisting of 3 repeats of the telomere sequence: TTAGGG or CCCTAA. Further, as noted earlier, the identification of large, ultrabright intranuclear foci in ≥1% of tumor cells is indicative of ALT-positivity. As with any FISH assay, appropriate positive and negative controls are important to identify potential staining issues and limit subjective interpretation issues. However, an internal telomere control to ascertain staining adequacy within a tissue specimen is unavailable. It should also be underscored that many ALT-negative neoplasms can demonstrate diminished telomere-specific FISH signals due to telomere attrition. As a result, significant technical expertise is required to ensure optimal telomere-specific FISH staining. Moreover, sufficient tumor cellularity with non-neoplastic cells for comparison is critical for analysis because of the low positive threshold for ALT (≥1%). The minimum number of tumor cells to evaluate by telomere-specific FISH has not been determined and, in the setting of limited biopsy specimens, can result in false-negative results. In a study of 13 EUS-FNA specimens, Hackeng et al. described a single case of an ALT-positive PanNET that was missed by telomere-specific FISH and attributed to inadequate sampling of the neoplasm [63]. Hence, further studies are necessary to establish appropriate testing parameters for telomere-specific FISH to reliably identify ALT-positive PanNETs.

9. Future testing strategies for ALT

Although telomere-specific FISH can be used to reliably assess for ALT, several novel technologies have recently been developed to identify ALT-positive tumors and can potentially address the challenges of FISH testing. In addition to large, ultrabright intranuclear telomeres, ALT is characterized by partially single-stranded telomeric (CCCTAA)_n DNA circles (C-circles) that are also ALT-specific [64]. The detection of C-circles can be evaluated from formalin-fixed paraffin-embedded tissues and, even, in blood specimens from patients with ALT-positive cancers [65]. The limitations of this assay are requirements for DNA amplification that often requires a radioactive label [66]. Although nonradioactive detection assays have been developed, the C-circle assay can be time-consuming and labor intensive.

With advancements in sequencing technologies, numerous studies of diverse tumor types have shown that the number of short reads containing telomeric repeats can be used to determine telomere content and length that are sufficient to detect ALT. Computational telomere content estimates have a good-to-excellent agreement with quantitative PCR-based measurements [67–69]. However, bioinformatic algorithms are significantly influenced by the sequencing protocol used, such as whole exome sequencing versus whole genome sequencing. Nevertheless, these DNA-based approaches coupled with the assessment of DAXX and ATRX, while requiring relatively little tumor amount, hold immense promise for future clinical use.

10. Conclusion

In summary, the status of DAXX, ATRX, and ALT can be clinically useful in multiple settings. For both preoperative and postoperative PanNET specimens, loss of DAXX/ATRX and the presence of ALT correlate with multiple adverse prognostic findings and are associated with distant metastases on follow-up. These biomarkers are especially important when evaluating patients with small (≤2.0 cm) PanNETs, which represents a treatment dilemma in determining whether continued surveillance or surgical management is appropriate. In contrast, among liver metastasis, patients with DAXX/ATRX-negative and/or ALT-positive PanNETs exhibit an improved overall survival rate than metastatic patients who harbor wild-type PanNETs. Considering genomic alterations in DAXX and ATRX, and subsequently ALT are largely unique to PanNETs, the assessment of DAXX/ATRX by immunohistochemistry and ALT by telomere-specific FISH are also invaluable diagnostic biomarkers. In the setting of a NET of unknown primary, DAXX/ATRX loss and ALT positivity would suggest a pancreatic origin. Similarly, for high-grade pancreatic neuroendocrine neoplasms, loss of expression for DAXX/ATRX and the presence of ALT would favor a diagnosis of a PanNET as opposed to a PanNEC. Considering the large breadth of independent studies that have confirmed the diagnostic and prognostic utility of DAXX, ATRX, and ALT, it is not surprising that these biomarkers have been clinically implemented in numerous laboratories around the world [70,71]. Consequently, forthcoming prospective studies are eagerly anticipated and will hopefully identify and address challenges with current testing methodologies.