Detection of Alternative Lengthening of Telomeres via Chromogenic in situ Hybridization (ALT-CISH) for the Prognostication of PanNETs and Other Neoplasms

Abstract

Molecular studies have shown ALT to be an important prognostic biomarker of shorter relapse-free survival (RFS) for patients with pancreatic neuroendocrine tumors (PanNETs) and other neoplasms. However, the preferred method of detecting ALT in tissue is by fluorescence in situ hybridization (FISH), which has several clinical limitations. These issues necessitate the creation of a chromogenic ALT assay that can be easily implemented into routine practice. A CISH assay was developed using genetically modified osteosarcoma cell lines, 20 normal pancreata, 20 ALT-positive PanNETs, and 20 ALT-negative PanNETs. Thereafter, it was validated on a multi-institutional cohort of 360 surgically resected PanNETs and correlated with multiple clinicopathologic features, RFS, and FISH results. Separately, 109 leiomyosarcomas (LMS) were evaluated by both CISH and FISH, and, similarly, the prognostic significance of ALT status was assessed. Upon optimization, ALT-CISH was identified in 112 of 360 (31%) primary PanNETs and was 100% concordant with FISH testing. ALT correlated with several adverse prognostic findings and distant metastasis (all p<0.004). The 5-year RFS for patients with ALT-positive PanNETs was 35% as compared to 94% for ALT-negative PanNETs. By multivariate analysis, ALT was an independent prognostic factor for shorter RFS. Similarly, ALT was associated with shorter RFS in LMS patients and, analogous to PanNETs, a negative, independent prognostic factor. ALT-CISH was developed and validated in not only PanNETs, but also sarcomas, specifically LMS. CISH testing has multiple advantages over FISH that facilitate its widespread clinical use in the detection of ALT and prognostication of patients with diverse neoplasms.

Introduction

Pancreatic neuroendocrine tumors (PanNETs) are the second most common neoplasm of the pancreas and are characterized by variable clinical behavior, pathologic findings, and molecular features. Historically, PanNETs were of low incidence, and patients often present with clinical signs and symptoms of peptide hormone hypersecretion (functional PanNETs).¹ With increasing use and new advancements in abdominal imaging, the incidence of PanNETs has increased rapidly and, currently, PanNETs affect 1 per 100,000 individuals annually.^{2, 3} This increasing incidence is mainly represented by a higher proportion of patients diagnosed with incidental, early-stage (localized) disease and an absence of hormonal hypersecretion syndrome (non-functional PanNETs). As many non-functional PanNETs are clinically indolent, the overtreatment of PanNETs has been a subject of debate, and, therefore, a surveillance approach may be warranted for a subset of patients.^4-6

To improve the surgical management of PanNET patients, the identification of prognostic biomarkers for PanNETs has been an area of active investigation. Existing prognostic biomarkers for patients with PanNETs are tumor size, World Health Organization (WHO) histologic grade, and emerging markers, such as the presence of growth patterns and tumor necrosis.^7-10 Numerous studies have shown that PanNETs of ≤2.0 cm (small) and classified as WHO grade 1 (low-grade) follow a benign clinical course with a very low risk of metastatic progression. However, several reports have found a subset of small, low-grade PanNETs can behave aggressively, including those initially predicted to be clinically indolent.^11-14 Moreover, the evaluation of tumor size and WHO grade are subjective in nature and do not accurately reflect the underlying biology of these neoplasms.^{15, 16}

Molecular studies have uncovered the genetic basis of PanNETs (i.e., recurrent somatic mutations in DAXX and ATRX) and have demonstrated metastatic PanNETs are frequently characterized by the activation of alternative lengthening of telomeres (ALT), a telomerase-independent and cancer-specific telomere maintenance mechanism.^17-21 Several investigative groups have reported the presence of ALT correlates with shorter relapse-free survival (RFS) for PanNET patients and is independent of tumor size and WHO grade.^{14, 22, 23} Furthermore, ALT is a prognostic biomarker for other neoplasms and is often associated with aggressive behavior in different types of sarcomas.²⁴ The preferred method of detecting ALT in tissue is telomere-specific fluorescence in situ hybridization (FISH), which defines the presence of ALT by the identification of large, ultrabright intranuclear telomeric foci in ≥1% of tumor cells.^{25, 26} However, as with any FISH assay, there are limitations to this methodology that preclude its adoption into clinical practice, such as associated costs and the requirement of specialized equipment and highly trained personnel. In contrast, chromogenic in situ hybridization (CISH) has a wider potential for implementation in routine pathology laboratories. The aims of this study were therefore to (1) develop an ALT-CISH assay for the assessment of PanNETs, (2) validate the prognostic utility of ALT using CISH in a large, multi-institutional cohort of PanNETs, and (3) determine the clinical utility of the CISH assay in other ALT-positive neoplasms, such as leiomyosarcomas (LMS).

Materials and Methods

Study population

Study approval was obtained from the University of Pittsburgh and Washington University Institutional Review Boards (IRB# PRO13020493). The surgical pathology archives from the Departments of Pathology at the University of Pittsburgh Medical Center (UPMC) and Barnes-Jewish Hospital were queried for neuroendocrine neoplasms of the pancreas between 1995 and 2020 that underwent surgical resection. Cases were cross-referenced with clinical and follow-up data obtained from patient electronic medical records. The study inclusion criteria consisted of the following: a solitary, well-differentiated neuroendocrine neoplasm [confirmed with positive immunolabeling for neuroendocrine markers (e.g., synaptophysin and chromogranin A)] centered within the pancreas, surveillance, and survival data of >2 years, absence of a genetic syndrome associated with pancreatic neuroendocrine neoplasms, and cases with sufficient material for ancillary studies (see Supplementary Materials and Methods).

Leiomyosarcomas (LMS) were searched from the UPMC Department of Pathology, Division of Anatomic Pathology, specimen archives between 1995 and 2020. Biopsy specimens were excluded, and only surgical resections were included if, upon cross-referencing with patient electronic medical records, performed for curative intent. All 109 cases were reviewed to confirm the diagnosis of LMS, and formally evaluated for site of origin, tumor differentiation, mitotic score, tumor necrosis, and FNCLCC (French Federation of Cancer Centers Sarcoma Group) grade. Pathology gross reports were assessed to identify tumor size and tumor site. Furthermore, follow-up data was obtained from the patient electronic medical records.

Chromogenic in situ hybridization (CISH)

Telomere-specific CISH was performed in the UPMC Clinical Immunohistochemical Laboratory, which is a CLIA-approved and CAP-accredited clinical laboratory. Creation and optimization of the telomere-specific CISH assay was performed on two autostainers: Leica BOND-III and Ventana Discovery ULTRA. For the Leica BOND-III, whole tumor sections and TMAs were cut at 4-um, deparaffinized, incubated for 5 min in BOND Epitope Retrieval Buffer (ER2) (Leica Biosystems) at 95°C, 5 min in enzyme protease at 40°C, and 10 min in H₂O₂ at room temperature. RNAscope (Advanced Cell Diagnostics) hybridization was performed for 120 min using the TELO-01zz probe (ready-to-use, catalog# 514038, search www.acdbio.com for “telo”) at 42°C, and followed by serial RNAscope diaminobenzene (DAB)-HRP amplification for 30 min and 15 min at 42°C. RNAscope amplification was performed three times and followed by 20 min in BOND Polymer Refine Detection at ambient temperature. Finally, a hematoxylin counterstain was performed for 10 min at ambient temperature (see Supplementary Materials and Methods for a step-by-step staining procedure).

The Ventana Discovery ULTRA was also used to compare the performance of telomere-specific CISH testing with the Leica BOND-III. Briefly, whole tumor sections were cut at 4-um, deparaffinized, underwent manufacturer recommendations for cell conditioning, followed by 8 min washing, and then 2 drops of the TELO-1zz probe (catalog# 514039) that is incubated for 4 minutes with the associated coverslip. RNAscope hybridization is done for 120 min at 43°C. Subsequent amplification of signal was performed using the RNAscope Amplification Kit (catalog #6614337001) for 32 min at 39°C and completed twice. This was followed by incubation for 12 min at ambient temperature and counterstained with hematoxylin for 8 min. Lastly, bluing reagent (catalog #5266769001) was applied and incubated for 8 min at ambient temperature. As the Ventana Discovery ULTRA is not an FDA-approved device, all telomere-specific CISH testing upon optimization of the assay was performed on the Leica BOND-III.

Scoring of ALT was performed by assessing at least 250 neoplastic nuclei for each case. Using previously described FISH criteria, ALT-positive cases were defined by the presence of large, dark intranuclear foci consistent with telomere signals in ≥1% of neoplastic nuclei, and the total signal intensity for individual foci >10-fold than telomere signals from non-neoplastic cells.^{14, 23}

Immunohistochemistry and FISH

In brief, immunohistochemical studies was performed on 4 um unstained slide sections from whole tumor sections or tissue microarrays (TMAs). Slides were deparaffinized with serial xylene treatments and subjected to antigen retrieval using heated citrate solution. Immunolabelling for ATRX (HPA001906 rabbit polyclonal, Sigma Aldrich) and DAXX (AB32140 rabbit monoclonal, Abcam) were performed on an automated Ventana Benchmark XT system using the biotin-free Ventana OptiView DAB IHC Detection Kit (Ventana Medical Systems).^{14, 23}

Telomere-specific FISH was performed as previously described using a fluorescently tagged telomeric-C PNA probe for whole tumor sections and TMAs.^{14, 23} A detailed description is provided within the Supplementary Materials and Methods. Scoring of ALT was performed by assessing at least 250 neoplastic nuclei for each case. Using previously described criteria, ALT-positive cases were defined by the presence of large, ultrabright intranuclear foci consistent with telomere signals in ≥1% of neoplastic nuclei, and the total signal intensity for individual foci >10-fold than telomere signals from non-neoplastic cells.

Statistical analysis

χ2 analysis or Fisher’s exact tests were used to compare categorical data, and Mann-Whitney U test was used to compare continuous variables. Survival curves were constructed using the Kaplan-Meier method and differences between groups were evaluated by the log-rank test. Recurrence-free survival (RFS) was calculated from the date of surgery to the date of first distant metastasis/local recurrence after surgery and censored at the date of last follow-up or death. Disease-specific survival (DSS) was calculated from the date of surgery to the date of death due to disease after surgery and censured at the date of last follow-up or death not due to disease. The prognostic significance of clinical and pathological characteristics was determined using univariate Cox regression analysis. Multivariate analyses of significant risk factors by univariate analysis were performed using Cox proportional hazard regression to identify independent risk factors for RFS and DSS. Inter-rater agreement with regards to the presence vs. absence of ALT of PanNETs and LMS between three pathologists was calculated using Fleiss’ kappa. All statistical analyses were performed using the SPSS Statistical software, V.28 (IBM) and statistical significance was defined as a p<0.05.

Results

Establishing a CISH assay for ALT

Osteosarcoma cell lines and whole tissue sections from 60 pancreatectomy specimens (20 histologically normal pancreata, 20 ALT-negative PanNETs, and 20 ALT-positive PanNETs) were used to establish an automated CISH assay for ALT (Supplementary Data and Supplementary Table 1). Briefly, ALT-CISH was first optimized to ensure the identification of weak, but distinct, punctate signals within the non-neoplastic nuclei of the islets of Langerhans and analogous to telomere-specific signals seen by FISH testing (Figure 1 and Supplementary Figure 1). Thereafter, ALT-negative, and ALT-positive PanNETs as defined by FISH were evaluated by CISH testing. In comparison to ALT-negative PanNETs, ALT-positive PanNETs were characterized by large, dark brown, and heterogeneous nuclear staining within ≥1% of neoplastic nuclei, but in all cases similar staining was observed in ≥10% of neoplastic nuclei. The ≥1% of neoplastic nuclei staining was used as criteria for the presence of ALT by CISH.

Figure 1.

Representative examples of normal pancreas (top row) and pancreatic neuroendocrine tumors (PanNETs) (middle and bottom rows) using fluorescence in situ hybridization (FISH) and chromogenic in situ hybridization (CISH) for detection of alternative lengthening of telomeres (ALT). (A) Normal pancreatic parenchyma with a benign islet of Langerhans (left) that shows barely perceptible telomere, nuclear (green) signals by telomere-specific FISH at low magnification (B) and high magnification (C). ALT-CISH exhibits weak, brown, intranuclear foci that is apparent at both low magnification (D) and high magnification (E). (F) For PanNETs that are negative for ALT, telomere-specific FISH testing had rare, weak telomeric signals within the neoplastic nuclei (G, low magnification [630x]; H, high magnification [1000x]). By ALT-CISH testing, the telomeric signals were less intense (I, low magnification [200x]; J, high magnification [400x]) than within normal islets of Langerhans, but still identifiable within the neoplastic nuclei. (K) ALT-positive PanNETs were characterized by large, ultrabright intranuclear telomeric foci in ≥1% of tumor cells that requires extensive assessment to ensure scoring of neoplastic nuclei and not non-neoplastic nuclei (L, low magnification; M, high magnification). In contrast, ALT-positive PanNETs stained using ALT-CISH were easily identifiable at low magnification (N) and high magnification (O) and using brightfield microscopy facilitated tumor correlation with corresponding H&E-stained slides.

Considering telomeric DNA sequences can be transcribed into long non-coding RNAs that are called telomeric repeat-containing RNA (TERRA) and may be detected by chromogenic in situ hybridization (CISH), genetically modified ALT-positive U2OS osteosarcoma cell lines with and without TERRA were generated and evaluated by ALT-CISH (Figure 2). As compared to control (scrambled) cells, knockdown of TERRA showed a loss in ALT-CISH signal intensity. Moreover, RNase A treatment of both control and TERRA knockdown U2OS cell lines showed similar telomeric signals by CISH with variable intensities present in ≥1% of neoplastic nuclei.

Figure 2.

U2OS osteosarcoma cells transfected with antisense oligonucleotides (ASO) to a control (scrambled) and TERRA sequence, and then evaluated by ALT-CISH. (A) U2OS cells with a control sequence (scrambled) show (B) large, dark-brown staining within the neoplastic nuclei. In comparison, (C) transfection with antisense oligonucleotides to TERRA showed depletion of (D) staining, but a residual punctate staining remained, likely consistent with telomeric signals. These findings were confirmed orthogonally with RNase A treatment (E through H) of both the control sequence (E and F) and antisense oligonucleotides to TERRA (G and H). Similarly, residual punctate staining remained, likely consistent with a telomere-specific signal.

Clinicopathologic assessment of surgically resected PanNETs using ALT-CISH

Both CISH and FISH were performed for 360 solitary, sporadic, and surgically resected PanNETs (Table 1). Based on CISH testing, ALT was detected in 112 (31%) PanNETs, was 100% concordant with ALT as defined by FISH, and exhibited marked intratumoral and intertumoral heterogeneity (Figure 3). ALT-positive PanNETs, as compared to ALT-negative PanNETs, were associated with a male predilection, larger tumor size, non-functional status, high WHO grade, lymphovascular invasion, perineural invasion, advanced pathologic T-stage, regional lymph node metastases, synchronous distant metastases, and metachronous distant metastases (all p<0.004). In addition, ALT correlated with loss of expression for DAXX/ATRX (p<0.001); however, 14 (13%) ALT-positive PanNETs had preserved expression for DAXX/ATRX.

Table 1. Clinicopathologic findings of 360 PanNET patients with respect to ALT status as defined by ALT-CISH testing.

Patient or Tumor Characteristics	Total (n = 360)	ALT-positive PanNET (n = 112, 31%)	ALT-negative PanNET (n = 248, 69%)	p
Gender
Female	177 (49%)	42 (38%)	135 (54%)	0.003
Male	183 (51%)	70 (62%)	113 (46%)
Median age (range), years	59.0 (22 - 93)	60.0 (27 - 93)	58.0 (22 - 85)	0.122
Median tumor size (range), cm	2.8 (0.6 - 18.0)	4.5 (1.0 - 18.0)	2.2 (0.6 - 17.0)	< 0.001
Location
Head, neck and uncinate	136 (38%)	39 (35%)	97 (39%)	0.482
Body and tail	224 (62%)	73 (65%)	151 (61%)
Functional PanNET	36 (10%)	3 (3%)	33 (13%)	0.001
WHO grade
Low (G1)	192 (53%)	29 (26%)	163 (66%)	< 0.001
Intermediate (G2)	160 (44%)	78 (70%)	82 (33%)
High (G3)	8 (3%)	5 (4%)	3 (1%)
Lymphovascular invasion	170 (47%)	89 (80%)	81 (33%)	< 0.001
Perineural invasion	110 (31%)	58 (52%)	52 (21%)	< 0.001
Primary tumor (pT) stage
pT1	104 (29%)	5 (4%)	99 (40%)	< 0.001
pT2	148 (41%)	48 (43%)	100 (40%)
pT3	106 (29%)	57 (52%)	49 (20%)
pT4	2 (1%)	2 (1%)	0 (0%)
Regional lymph node (pN) stage*
pN0	183 (59%)	41 (37%)	142 (71%)	< 0.001
pN1	127 (41%)	69 (63%)	58 (29%)
Synchronous distant metastasis	68 (19%)	40 (36%)	28 (11%)	< 0.001
Metachronous distant metastasis/recurrence**	58 (20%)	42 (58%)	16 (7%)	< 0.001
DAXX/ATRX aberrant (loss of) expression	101 (28%)	98 (87%)	3 (1%)	< 0.001
ALT (presence of) based on telomere-specific FISH	112 (31%)	112 (100%)	0 (0%)	< 0.001

Abbreviations: ALT, alternative lengthening of telomeres; CISH, chromogenic in situ hybridization; FISH, fluorescence in situ hybridization; PanNET, pancreatic neuroendocrine tumor; WHO, World Health Organization

^*Data regarding regional lymph node status was available for 310 (86%) PanNET patients.

^**Excludes 68 (19%) PanNET patients with synchronous distant metastasis.

Figure 3.

Representative examples of ALT-positive PanNETs (A, C, E, G, I, and K, H&E-stained) that were evaluated by ALT-CISH (B, D, F, H, J, and L). CISH testing of ALT-positive PanNETs showed wide variability in staining that was characterized by marked intratumoral and intertumoral heterogeneity of telomeric foci. However, in all ALT-positive cases, large, dark brown nuclear staining within ≥1% of neoplastic nuclei was observed.

Prognostic significance of ALT-positive PanNETs by CISH

Patients whose tumors demonstrated ALT positivity had shorter relapse-free survival (RFS) and shorter disease-specific survival (DSS) than patients with ALT-negative PanNETs (Figure 4). Among ALT-positive PanNET patients, the RFS rates were 84%, 35%, and 32% at 1, 5, and 10 years, respectively. The DSS rates were 82% at 5 years and 54% at 10 years. In comparison, patients with ALT-negative PanNETs had significantly longer RFS (99% at 1 year, 94%, at 5 years, and 89% at 10 years, p<0.001) and improved DSS (92% at 5 years, and 84% at 10 years, p<0.001).

Figure 4.

Kaplan-Meier curves comparing relapse-free survival (RFS) and disease-specific survival (DSS) after surgical resection among PanNET patients with respect to ALT status by ALT-CISH testing. Patients with ALT-positive PanNETs have shorter RFS (A) and shorter DSS (B), as compared to patients with ALT-negative PanNETs.

Results of univariate and multivariate Cox regression analysis for RFS and DSS in relation to various clinicopathologic features including ALT status by CISH are shown in Table 2. By univariate analysis, shorter RFS and shorter DSS were associated with patient age as a continuous variable (p=0.005 and p=0.007), tumor size >2.0 cm (p<0.001 for both), G2-to-G3 WHO grade (p<0.001 for both), lymphovascular invasion (p<0.001 for both), perineural invasion (p<0.001 for both), advanced tumor stage (pT3 and pT4 vs. pT1 and pT2) (p<0.001 for both), regional lymph node metastases (p<0.001 for both), loss of DAXX/ATRX expression by immunohistochemistry (p<0.001 for both), and the presence of ALT by CISH (p<0.001 for both). Multivariate analysis was used to determine the prognostic significance of ALT based on CISH for RFS and DSS, and included patient age, tumor size >2.0 cm, WHO grade, lymphovascular invasion, perineural invasion, and regional lymph node metastases. The presence of ALT by CISH was an independent prognostic factor for shorter RFS (HR: 4.458 [2.254-8.819], p<0.001) but not for shorter DSS (p=0.845).

Table 2. Univariate and multivariate Cox regression analyses for relapse-free survival^* and disease-specific survival among 360 PanNET patients.

	Univariate Cox Regression Analysis				Multivariate Cox Regression Analysis
Patient or Tumor Characteristics	Relapse-Free Survival HR (95% CI)	p	Disease-Specific Survival HR (95% CI)	p	Relapse-Free Survival HR (95% CI)	p	Disease-Specific Survival HR (95% CI)	p
Gender, female vs. male (ref.)	0.840 (0.500 - 1.411)	0.510	0.893 (0.507 - 1.574)	0.696
Age, years**	1.029 (1.009 - 1.051)	0.005	1.031 (1.009 - 1.054)	0.007	1.008 (0.988 - 1.028)	0.439	1.020 (0.998 - 1.043)	0.080
Tumor size, >2.0 cm vs. ≤2.0 cm (ref.)	15.115 (4.727 - 48.335)	<0.001	26.176 (3.611 - 189.761)	<0.001	7.158 (1.703 - 30.091)	0.007	7.889 (1.051 - 59.199)	0.045
Location, head, neck and uncinate vs. body and tail (ref.)	1.184 (0.692 - 2.024)	0.538	1.592 (0.904 - 2.804)	0.108
WHO grade, G2 or G3 vs. G1 (ref.)	4.953 (2.797 - 8.770)	<0.001	5.302 (2.684 - 10.473)	<0.001	1.588 (0.810 - 3.113)	0.178	2.301 (1.057 - 5.009)	0.036
Lymphovascular invasion, presence vs. absence (ref.)	6.243 (3.483 - 11.192)	<0.001	8.274 (3.784 - 18.090	<0.001	1.055 (0.519 - 2.143)	0.883	2.002 (0.798 - 5.023)	0.139
Perineural invasion, presence vs. absence (ref.)	5.789 (3.392 - 9.880)	<0.001	3.521 (1.972 - 6.288)	<0.001	1.529 (0.818 - 2.858)	0.184	0.871 (0.463 - 1.640)	0.871
Tumor stage (pT), pT3 and pT4 vs. pT1 and pT2 (ref.)	3.863 (2.300 - 6.488)	<0.001	3.062 (1.728 - 5.426)	<0.001
Lymph node metastasis (pN), pN1 vs. pN0 (ref.)***	5.651 (3.231 - 9.885)	<0.001	4.901 (2.547 - 9.431)	<0.001	2.196 (1.133 - 4.257)	0.020	2.143 (1.027 - 4.472)	0.042
DAXX/ATRX expression, aberrant vs. non-aberrant (ref.)	12.319 (6.889 - 22.032)	<0.001	3.330 (1.878 - 5.906)	<0.001
ALT by ALT-CISH, presence vs. absence (ref.)	13.594 (7.416 - 24.917)	<0.001	3.271 (1.837 - 5.824)	<0.001	4.458 (2.254 - 8.819)	<0.001	1.063 (0.575 - 1.964)	0.845

Abbreviations: ALT, alternative lengthening of telomeres; CISH, chromogenic in situ hybridization; DSS, disease-specific survival; FISH, fluorescence in situ hybridization; PanNET, pancreatic neuroendocrine tumor; ref., reference; RFS, relapse-free survival

^*Cox regression analysis is based on data available for 292 (81%) PanNET patients.

^**The patient or tumor characteristic of interest was evaluated as a continous variable.

^***Cox regression analyses for RFS and DSS are based on data available for 244 (68%) and 310 (86%) PanNET patients, respectively.

ALT-CISH testing of surgically resected leiomyosarcomas

In addition to PanNETs, previous studies have also found the presence of ALT to be prognostic for patients with leiomyosarcomas (LMS). Therefore, to examine if the clinical utility of ALT-CISH can be expanded to other neoplasms, a cohort of 109 surgically resected LMS from multiple sites of origin were evaluated with CISH and FISH (Figure 5). Based on CISH, ALT was detected in 44 (40%) LMS and was 100% concordant with ALT as defined by FISH (Supplementary Table 2). The presence of ALT by CISH in LMS was associated with larger tumor size, synchronous distant metastases, metachronous distant metastases/recurrences, and loss of expression for ATRX (all p<0.004). Moreover, 4 (9%) ALT-positive LMS had preserved expression for ATRX.

Figure 5.

Representative examples of leiomyosarcomas (LMS), which were evaluated by both FISH and CISH. (A) An ALT-negative LMS with weak (green), intranuclear (telomeric) signals that show little variability in size and number by FISH testing (B). Similarly, ALT-CISH shows weak, brown, punctate signals within the neoplastic nuclei that are identifiable at low magnification (C). (D) Within an ALT-positive LMS, the telomere signals are large, ultrabright, and variable in size by FISH (E). ALT-CISH testing also shows large, dark brown, and variable sized signals; however, staining is easily perceptible at low magnification, which facilitates histologic correlation.

Similar to PanNETs, patients with ALT-positive LMS had shorter RFS and shorter DSS than patients with ALT-negative LMS (Figure 6). The RFS rates for ALT-positive LMS patients were 60% at 1-year and 20% at 3-years, while RFS rates for ALT-negative LMS patients were 80% at 1-year and 53% at 3-years. Further, the DSS rates for ALT-positive LMS patients were 55% at 3-years and 40% at 5-years, while ALT-negative LMS patients were 75% at 3-years and 68% at 5-years. By univariate analysis (Supplementary Table 3), shorter RFS and shorter DSS were associated with tumor size as a continuous variable (p=0.033 and p<0.001), mitotic activity of >11 per 1 mm² (p=0.009 and p=0.011), FNCLCC grade 3 (p=0.042 and p=0.012), and the presence of ALT by CISH (p=0.005 and p=0.004). Tumor necrosis of ≥50% was associated with shorter DSS (p=0.013), but not shorter RFS (p=0.336). Finally, by multivariate analysis using a model to include tumor size, mitotic activity, FNCLCC grade, and ALT, the presence of ALT by CISH was an independent prognostic factor for shorter RFS (HR: 1.737 [1.011-2.982], p=0.043) but not shorter DSS (p=0.078).

Figure 6.

Kaplan-Meier curves comparing RFS and DSS after surgical resection among LMS patients with respect to ALT status by ALT-CISH testing. Patients with ALT-positive LMS have shorter RFS (A) and shorter DSS (B), as compared to patients with ALT-negative LMS.

Pathologist agreement in interpretation of ALT based on ALT-CISH

In addition to evaluating the concordance between CISH and FISH, the level of agreement between pathologists in interpreting the results for ALT-CISH assay was determined using a cohort of 50 PanNETs (25 ALT-positive and 25 ALT-negative by FISH) and 50 LMS (25 ALT-positive and 25 ALT-negative by FISH). Three pathologists (designated A, B, and C) independently scored each CISH-stained case as positive or negative for ALT. Among the 50 PanNETs, there was a perfect correlation between Pathologist A and Pathologist C (κ=1.000), while Pathologist B had one disagreement (κ=0.961), interpreting an ALT-positive PanNET as ALT-negative by CISH (Supplementary Figure 2A). By comparison, for 50 LMS cases, there was a perfect correlation between Pathologic A and Pathologist B (κ=1.000), while Pathologist C had two disagreements (κ=0.923), interpreting two ALT-positive LMS as ALT-negative by CISH (Supplementary Figure 2B). A re-review of the three discordant cases amongst all three pathologists revealed intratumoral heterogeneity of ALT as the source of discordance in interpretation. Upon education of ALT heterogeneity within both PanNETs and LMS, random re-review yielded no discordance and perfect recognition of ALT-positive vs. ALT-negative cases.

Discussion

The clinical rationale for determining the ALT status of PanNETs lies in its ability to convey important patient prognostic information. Herein, we developed a novel CISH assay for the detection of ALT in PanNETs. The presence of ALT based on CISH was detected in 31% of surgically resected, primary PanNETs and associated with several adverse prognostic features to include larger tumor size, high WHO grade, lymphovascular invasion, perineural invasion, advanced pathologic T-stage, regional lymph node metastases, synchronous distant metastases, and metachronous distant metastases. Moreover, the identification of ALT by CISH was a prognostic biomarker for shorter RFS for PanNET patients and independent of tumor size, WHO grade, lymphovascular invasion, perineural invasion, and regional lymph node metastases. As further proof to the clinical applicability of ALT-CISH, the identification of ALT positivity in LMS was also associated with multiple adverse prognostic features and correlated with shorter RFS in LMS patients. In fact, analogous to PanNETs, the presence of ALT in LMS was an independent prognostic factor for shorter RFS.

Our findings are consistent with previous studies that have reported the presence of ALT based on telomere-specific FISH predicts a poor prognosis for patients with a PanNET. The first description of ALT as an adverse molecular feature for PanNETs was by Marinoni et al., who showed a correlation between ALT and reduced RFS.²⁷ These observations were further confirmed by multiple, multi-institutional studies.^{14, 19-23, 28} Kim et al. and Singhi et al. highlighted a statistically significant association between ALT and tumor size, perineural and lymphovascular invasion, nodal metastases, and distant metastases.^{22, 23} Both investigative groups found ALT positivity correlated with higher WHO grade, but ALT was also present in low-grade (WHO grade 1) tumors. In particular, Scarpa et al. described the presence of ALT in WHO grade 1 and WHO grade 2 PanNETs and its association with chromosomal instability, which has been hypothesized to be a key event in the molecular pathogenesis of metastatic disease in PanNET patients.¹⁸ The seminal manuscript demonstrating the clinical utility of ALT was by Hackeng et al., which focused on an analysis of a large, international cohort of non-functional PanNETs.¹⁴ The authors established ALT as a negative, independent prognostic biomarker for RFS including patients with small PanNETs (≤2.0 cm), which often represent a clinical dilemma for surgical management.

It is important to underscore that the detection of ALT by CISH was 100% concordant with FISH testing. Although FISH is the gold standard assay for ALT assessment in archived tissue, this technique presents several disadvantages. FISH testing requires costly equipment and specialized personnel for signal detection and recognition.^{29, 30} Fluorescence microscopes equipped with high-magnification oil immersion objectives and multiband filters are indispensable but are often too expensive for most diagnostic laboratories. As fluorescence signals fade within a few weeks, hybridization results should ideally be recorded with a high-priced charged-coupled device camera. Since distinguishing between neoplastic and non-neoplastic cells can be complicated and tedious to analyze under fluorescence, evaluation of FISH signals requires experienced and well-trained staff for interpretation of tumor specimens.^{31, 32} Nevertheless, some authors suggest that because FISH testing is performed at high magnification, intratumoral heterogeneity for ALT can be missed.^{23, 33}

As an alternative to telomere-specific FISH, several technologies have been developed to identify ALT-positive tumors. For instance, ALT is not only characterized by large, ultrabright intranuclear telomeric signals, but also by partially single stranded telomeric DNA circles (C-circles).^{34, 35} C-circle analysis however requires multiple analytic steps that include DNA amplification using a radioactive label, which are typically unavailable to many laboratories.³⁶ While non-radioactive detection assays for C-circles have been developed, these techniques are laborious and cost prohibitive as a routine clinical test. Another option for evaluating ALT is the assessment of DAXX and ATRX status by immunohistochemical expression. For PanNETs and LMS, ALT positivity is reported to correlate with protein loss of DAXX and/or ATRX.^{14, 22, 23, 27, 37, 38} Similarly, our study found 87% of ALT-positive PanNETs had DAXX/ATRX loss and 91% of ALT-positive LMS had ATRX loss. However, loss of DAXX/ATRX has also been documented to occur in the absence of ALT positivity. Furthermore, some ALT-positive tumors (e.g., hepatocellular carcinomas) do not exhibit loss of DAXX/ATRX.³⁹ Thus, while statistically significant, the association between ALT and DAXX/ATRX loss is imperfect and, therefore, a quick and cost-effective approach for the detection of ALT is needed.

In contrast to FISH, most pathology laboratories can easily implement ALT-CISH testing as it requires routinely available equipment, such as an FDA-approved chromogenic autostainer and a conventional brightfield microscope. In general, chromogenic probes for nucleic acid sequences are less expensive than fluorescence-labeled probes.⁴⁰ Additionally, since telomere results can be visualized using a standard microscope, CISH interpretation allows for the simultaneous evaluation of neoplastic morphology and telomere signals, reducing the risk of analyzing the nonmalignant component.⁴¹ Another advantage of CISH is tumor heterogeneity can be promptly recognized, even at low magnification. Since the CISH assay allows for easier identification of ALT at low magnification, analysis of CISH signals is consequently less time-consuming than FISH testing. Chromogenic signals are permanent and, therefore, slides can be stored and used for reevaluation, training, and easily digitized for image-based analysis.

In addition to its prognostic utility, the detection of ALT by CISH has potential treatment ramifications. ALT-positive cells have characteristically high levels of replication stress that perpetuate ALT-mediated homology-directed repair.^42-44 ATM contributes to the replication stress response and inhibitors of ATM may exacerbate telomere replication stress and cell death in ALT-positive tumors. Specifically, the ATM inhibitor AZD0156 has shown selective toxicity in ALT-positive neuroblastomas and is currently in phase I clinical trials.⁴⁵ Furthermore, resistance to the chemotherapeutic drugs temozolomide and SN38 (the active metabolite of irinotecan) used upon relapse was reversed when neuroblastoma cell lines were treated with AZD0156. The ATR kinase is recruited to sites of replication stress, and it was previously shown that ATR inhibitors had greater toxicity in ALT-positive neoplasms as compared to telomerase-mutant neoplasms.⁴⁶ There are currently multiple ATR inhibitors at various stages of clinical development, including phase I (BAY1895344 and M4344) and phase II (M6620 and AZD6738) clinical trials.^47-50

It is important to note that telomere DNA sequences at chromosomal ends are transcribed into long non-coding RNAs that are known at telomeric repeat-containing RNA (TERRA). In ALT-positive neoplastic cells, TERRA is commonly upregulated.^{46, 51} Furthermore, TERRA forms R-loops (a DNA secondary structure of three nucleic acid strands that contain a DNA:RNA hybrid and a displaced DNA strand) at telomeres, which contribute to maintenance of ALT-positive telomeres.^52-55 Considering TERRA consists of telomeric DNA sequences, it is not surprising that ALT-CISH using an RNA-based probe hybridizes to not only telomeric ends of chromosomes, but also TERRA within neoplastic cells. Moreover, in the setting of ALT, it is expected that the CISH signals will be amplified with high expression of TERRA. This hypothesis is supported by our finding that telomere signals in U2OS osteosarcoma cells (an established ALT-positive cell line) were decreased upon knockdown of TERRA. Furthermore, RNase A treatment of U2OS cells with and without TERRA showed similar results. Nevertheless, heterogeneous telomeric signals in ≥1% of neoplastic nuclei that were consistent with the presence of ALT were identified upon TERRA knockdown and/or RNase treatment and analogous to findings commonly seen with telomere-specific FISH.

There are several limitations to our study that should be acknowledged. It is retrospective in design and although a large cohort of PanNETs and LMS were evaluated, this study may suffer from a surgical selection bias. In addition, not all patients received the same form of treatment. While a pancreaticoduodenectomy or distal pancreatectomy was performed in the majority of PanNET cases, 14% of patients underwent an enucleation or central pancreatectomy, and thus, regional lymph node sampling may be inadequate. Removing these patients from our analysis would have little impact on the statistical associations and prognostic findings of ALT by CISH. Of note, enucleation and central pancreatectomy procedures are typically done in the setting of small PanNETs (≤2.0 cm) because these tumors often have an indolent natural history. However, multiple studies have reported that a subset of small PanNETs can pursue an aggressive clinical course. Within our cohort, 7% of ALT-positive PanNETs were ≤2.0 cm in size and, among these small ALT-positive PanNETs, 33% developed distant metastatic disease. Another point of contention is TMAs were used to validate CISH testing of PanNETs and LMS. TMAs may not be representative of the entire tumor as compared to whole tissue sections and can be especially challenging in the setting of tumor heterogeneity with telomere-specific FISH. The issue of tumor heterogeneity however is minimized considering the multiplicity of cores for each tumor within the TMAs and the use of CISH as opposed to FISH. Admittingly, the lack of pathologist recognition of ALT intratumoral heterogeneity may be a source of error for the CISH assay. For both PanNETs and LMS, there was pathologist disagreement; however, this occurred in a small minority of cases and upon understanding the presence of intratumoral heterogeneity of ALT, the level of agreement on re-review was perfect. Additionally, ALT-CISH testing may not solely identify elongated telomeres. As mentioned previously, ALT is associated with both elongation of telomeric DNA sequences and the upregulation of TERRA, which includes telomeric repeats. A significant disadvantage of telomere-specific FISH testing is the relative paucity of ultrabright fluorescent signals to identify variability and lengthening of telomeres. In comparison to telomere-specific FISH, the signal intensity of CISH is the setting of ALT is markedly increased due to its ability to detect elongated telomeres and increased TERRA levels. Thus, cross hybridization with TERRA by ALT-CISH represents an advantage rather than a disadvantage in the assessment of ALT status. Finally, a formal cost analysis of CISH versus FISH testing was not performed within our study. Unfortunately, a calculation of true cost is difficult to determine as it must account for a diverse number of parameters, such as personnel, space, equipment, reagents, specimen volume, depreciation, and shared resources (e.g., autostainers). Nevertheless, the charges associated with performing CISH are well-known to be cheaper than those for FISH testing, and solely considering the purchase of a ready-to-use probe from Advanced Cell Diagnostics for either a Leica or Ventana autostainer, the cost was ~$15 per slide.

In summary, we report the creation and validation of a CISH assay for the evaluation of ALT status in not only PanNETs, but also other tumor types, such as LMS. This assay was completely concordant with the gold standard assay for ALT detection in tissue, telomere-specific FISH, and correlated with several adverse prognostic features including RFS for both PanNETs and LMS. In comparison to FISH testing, CISH has multiple advantages that include decreased cost, quicker turnaround time, and accurate interpretation of neoplastic cells with traditional brightfield microscopy. While further studies are required to assess the utility of ALT-CISH testing in clinical practice, the automated nature of this assay facilitates its implementation in a wide number of pathology laboratories.

Key Abbreviations:

ALT

alternative lengthening to telomeres

CISH

chromogenic in situ hybridization

DSS

disease-specific survival

FISH

fluorescence in situ hybridization

LMS

leiomyosarcoma

PanNET

pancreatic neuroendocrine tumor

RFS

relapse-free survival

Data Availability

The protocols generated in this study are outlined within the manuscript and supplementary files. However, additional details including the autostainer programs, reagents, and conditions are freely available upon request.