Role of CDKN2C Fluorescence In Situ Hybridization in the Management of Medullary Thyroid Carcinoma

Abstract

Medullary thyroid carcinoma (MTC), an aggressive form of thyroid cancer, occurs sporadically in approximately 75% of MTCs. RET and RAS mutations play a role in about 40% and 15%, respectively, of sporadic MTCs and are predominant drivers in MTC pathways. These mutations are some of the most comprehensively described and screened for in MTC patients; however, in recent studies, other mutations in the CDKN2C gene (p18) have been implicated in the tumorigenesis of MTC. Comparative genomic hybridization analysis revealed that approximately 40% of sporadic MTC samples have loss of CDKN2C at chromosome 1p32 in addition to frequent losses of CDKN2D (p19) at chromosome 19p13. However, no feasible routine method had been established to detect loss of heterozygosity (LOH) of CDKN2C and CDKN2D. The aim of this study is to assess the feasibility of using Fluorescence in situ Hybridization (FISH) to screen MTC patients for CDKN2C and CDKN2D deletions. We subjected 5 formalin-fixed, paraffin-embedded (FFPE) MTC samples with defined RET/RAS mutations to dual-color FISH assays to detect loss of CDKN2C and/or CDKN2D. We prepared spectrum orange probes using the bacterial artificial chromosomes RP11–779F9 for CDKN2C (p18) and RP11–177J4 for CDKN2D (p19) and prepared spectrum green control probes to the 1q25.2 and 19q11 regions (RP11–1146A3 and RP11–942P7, respectively). Nine FFPE normal thyroid tissue samples were used to establish the cutoff values for the FISH signal patterns. Of the five FFPE MTC samples, four and one yielded a positive significant result for CDKNN2C loss and CDKN2D loss, respectively. The results of a Clinical Laboratory Improvement Amendments validation with a CDKN2C/CKS1B probe set for CDKN2C (p18) loss of heterozygosity were 100% concordant with the FISH results obtained in this study. Thus, FISH is a fast and reliable diagnostic or prognostic indicator of gene loss in MTC.

Introduction

Medullary thyroid carcinoma (MTC), which arises from the calcitonin-producing parafollicular C cells of the thyroid [1], accounts for 5–10% of all thyroid cancers. Hereditary, MEN2-associated MTC accounts for approximately 25% of MTC cases, whereas sporadic MTC accounts for 75% of cases [2,3]. The predisposition to developing hereditary MTC was mapped to chromosome 10 by genetic linkage analysis in 1987; and germline mutations of RET, causing constitutive activation of the proto-oncogene, were found to be responsible for the disease in 1993 [4–6]. Underscoring the importance of RET in MTC, the U.S. Food and Drug Administration recently approved the use of the molecular targeted therapies Vandetanib and Cabozantinib, both of which inhibit the action of RET [7], for the treatment of MTC. Although RET mutations are the major drivers of sporadic MTC, they occur in only about 40% of sporadic MTCs [8]. In addition, mutations in RAS had been implicated in approximately 10–15% of sporadic cases, labeling it as a second driver of MTC [9]. These mutations, even though they are predominant drivers of MTCs, still do not account for all cases of sporadic MTC and its tumorigenesis.

To date, no other major drivers of C-cell oncogenesis have been identified. An array-based comparative genomic hybridization study first performed by Ye et al. revealed many copy number alterations in MTC, indicating that C-cell oncogenesis may be driven by copy number alterations rather than gene mutations [10]. Other studies found increased MTC risk to be associated with cell cycle genes, including those in the cyclin D–Cdk4–6/INK4/Rb/E2F pathway, which integrates mitogenic and antimitogenic stimuli to control cell growth [11,12]. Cyclin-dependent kinase inhibitor (CDKN) family members, which include p15, p16, p18, and p19, play a crucial role in preventing inappropriate cell division by binding to CDK4 or CDK6 and inhibiting the action of cyclin D, leading to G1 cycle arrest. Frequently mutated or deleted in a wide variety of tumors, CDKN genes increase RB phosphorylation, which in turn disrupts the inhibitory RB-E2F interaction. The free E2F transcribes genes and then drives the cell through S phase, suggesting that the CDKN family members have important functions as tumor suppressors [13].

Loss of the p18 gene, CDKN2C, has been associated with RET-mediated MTC [14–17]. Additionally, the frequent loss of the 19p13 region in MTC implicates the p19 gene, CDKN2D, though the exact nature of that role remains unclear [10,17]. Those genes involved in the CDKN pathway, such as CDKN2C and CDKN2D, loss of heterozygosity (LOH) suggests that MTC patients with RET/RAS mutations have additional molecular abnormalities. This is supported by a more detailed analysis of available array comparative genomic hybridization studies [10,17]. In the MD Anderson cancer center data set, which included 20 sporadic MTCs, 14 (70%) had LOH in regions containing either CDKN2C or CDKN2D. Furthermore, an independent analysis of 41 sporadic MTC tumors revealed loss of heterozygosity (LOH) in CDKN2C, at chromosome 1p32 and CDKN2D, at chromosome 19p13 for 39% and 24% of cases, respectively [17]. These findings suggest that CDKN2C and CDKN2D loss need to be routinely screened for in patients with sporadic MTCs and eventually correlated with clinical outcomes to provide a more rational approach to selecting individualized therapies targeting CDK inhibition in MTC of which current research on CDK4 inhibitors for various cancer therapies is ongoing [18–20].

Even though CDKN2C and CDKN2D LOH had been shown to be highly implicated in sporadic MTC cases, no feasible routine method had been established to assess for those losses. The aim of the present study is to assess the feasibility of using fluorescence in situ hybridization (FISH) to screen MTC patients for CDKN2C and CDKN2D deletions.

Materials and Methods

Probe Design and Preparation.

In FISH, the probes are fluorescently labeled single stranded DNA that are complementary to the targeted genomic sequence. In order to attain those DNA fragments complementary to our regions of interest, we searched the University of California Santa Cruz Genome Browser (http://genome.ucsc.edu/cgi-bin/hgGateway) to identify bacterial artificial chromosome (BAC) clones (Figure 1). We ordered the BAC clones RP11–779F9 (for CDKN2C) and RP11–177J4 (for CDKN2D), along with their respective controls RP11–1146A3 (located on 1q25.2) and RP11–942P7 (located on 19q11) from http://bacpac.chori.org/. Bacterial stocks were plated and cultured overnight at 37°C and a single colony was then inoculated in 25 ml of Luria-Bertani broth with 25 μg/ml chloramphenicol. The culture was placed in a controlled environment incubator shaker set to 260 rpm at 37°C for 8–12 hours. DNA extraction was then performed using the Purelink HiPure Plasmid Miniprep kit (Invitrogen) according to the manufacturer’s instructions. A Vysis Nick Translation kit (Abbott Laboratories, Abbott Park, IL) was used according to the manufacturer’s instructions to directly label BAC DNA with SpectrumOrangefortargetedregionsandSpectrumGreen for control regions. Working solutions for each of the BAC probes were hybridized on blood slides for validation. The validation included hybridizing the probes on blood slide metaphases followed by reverse DAPI through Cytovision AI imaging system to confirm the proper position of the probes on chromosome 1 and 19 with the absence of non-specific or specific hybridization on other chromosomes’ regions.

Figure 1.

We searched the University of California Santa Cruz Genome Browser to identify BAC clones. The BAC clone RP11–779F9 spans CDKN2C on chromosome 1p; the control BAC clone RP11–1146A3 is on the q-arm. The BAC clone RP11–177J4 spans CDKN2D on chromosome 19p; the control BAC clone RP11–932P7 is on the q-arm.

Pretreatment of Formalin-Fixed, Paraffin-Embedded Tissues.

Nine slides of formalin-fixed, paraffin-embedded (FFPE) normal thyroid tissue and five slides of FFPE MTC with defined RET/RAS mutations derived from patients who were treated at The University of Texas MD Anderson Cancer Center were randomly selected. All cases were derived from patients who were treated at The University of Texas MD Anderson Cancer Center. Pathologically confirmed normal and medullary thyroid carcinoma tumor tissue was obtained with the approval of The University of Texas MD Anderson Cancer Center’s Institutional Review Board. The slides were initially deparaffinized by placing them in the oven at 65°C overnight and then washing them in a series of CitriSolv. The slides were then pretreated with the Spot-LightTissue Pretreatment Kit (Invitrogen). Briefly, the slides were dehydrated by serial washes of 100% ethanol and then placed in pretreatment solution for 33 minutes at 95–98°C. The slides were then washed with buffer (2x saline sodium citrate [SSC]/0.025% Tween 20) several times and dehydrated with serial up-graded washes of ethanol (50%, 70%, 90%, and 100%). Tumor borders were delineated on each slide by using a DAKO cytomation pen. The slides were then incubated with the enzyme reagent for 30 minutes at 37°C, washed several times with buffer, and finally subjected to serial up-graded washes of ethanol.

Fluorescence in Situ Hybridization.

CDKN2C and/or CDKN2D loss were detected by a dual-color FISH assay. The probes were initially denatured at 75°C for 10 minutes to improve their hybridization. The CDKN2C and CDKN2D probes with their respective controls were added to the pretreated FFPE slides. The slides were coverslipped, allowing the probes to disperse, and then sealed with rubber cement. Probe and slide are co-denatured at 85°C for 10 minutes and hybridization at 37°C for 24–48 hours in a Thermobrite machine. The slides were washed in 2xSSC for 2 minutes each followed by washes in 2x SSC/0.025% Tween at 65°C for 5–10 minutes and 2SSC/0.025% Tween at room temperature for 2 minutes. The slides were subjected to washes of 2xSSC and dehydrated via serial up-graded washes of ethanol (70%, 95%, and 100%). The slides were air-dried subjected to serial CitriSolv washes and then subjected to two 100% ethanol washes for 5 minutes each step. Approximately 20 μl of room-temperature DAPI was applied to the hybridization area. The slides were cover slipped and stored in darkness for 20 minutes at 4°C and then examined under a fluorescence microscope.

FISH in CLIA Laboratory.

For the cytogenetic Clinical Laboratory Improvement Amendments (CLIA) test, the commercial probe set developed by Vysis for detecting CDKN2C loss in multiple myeloma was utilized to expand the FISH analysis in clinical settings [21]. We used the CDKN2C/CKS1B probe set, which consists of a red-labeled 180-kbp probe that spans the entire CKS1B gene and a green-labeled 168-kbp probe that spans the entire CDKN2C gene. The initial validation determines its clinical sensitivity and specificity as well as gathering the normal reference range and cut off for its future applicability in the clinical setting.

Statistical Analysis.

Dual-color FISH data were determined to be significant using a cutoff value established with the BETAINV statistical tool through Microsoft Excel. The cutoff for FISH analysis is usually conducted from the analysis of 200 cells, 100 cells by two different technologists. This is performed by identifying the normal case with the highest number of false positive nuclei for any given signal pattern and inserted into the BETAINV function with a 95% confidence interval [22].

Results

In a validation test with metaphases from human blood, developed homebrew FISH probes for CDKN2C and CDKN2D was hybridized at chromosome 1p32 and chromosome 19p13, respectively (Figure 2). Among nine selected normal thyroid FFPE samples, all had a normal 2R2G FISH signal pattern (Figure 3). The cutoff values obtained with the BETAINV statistical tool for different signal expression patterns are given in Table 1. The CDKN2C gene had a cut off of 10% and 12 % for 1R2G and 1R1G signal pattern, respectively. The CDKN2D gene, on the other hand, had a cut off of 9.5% and 10.5% for 1R2G and 1R1G signal pattern, respectively. Of the five FFPE MTC samples with defined RET/RAS mutations, four and one yielded a positive significant result for CDKN2C loss and CDKN2D loss, respectively (Table 2). The results of the cytogenetic CLIA test were 100% concordant with our FISH results (Table 3).

Figure 2.

Validation of the homebrew probes for the CDKN2C and CDKN2D genes and their respective control probes. (A) The probe for CDKN2C, located at 1p32, is shown in the 1p region (red), and its control probe is shown in the 1q region (green). (B) The probe for CDKN2D, located at 19p13, is shown in the 19p region (red), and the control probe is shown in the 19q region (green).

Figure 3.

Normal thyroid tissue showed an overall normal FISH 2R2G signal pattern. MTC tissue showed an abnormal FISH 1R2G signal pattern, indicating a deletion in the 1p32 region.

Table 1.

Cutoff values for the CDKN2C and CDKN2D genes for the 1R2G and 1R1G fluorescence in situ hybridization signal patterns.

Gene	1R2G (%)	1R1G (%)
CDKN2C	10.0	12.0
CDKN2D	9.5	10.5

Table 2.

Percentages of 1R2G and 1R1G fluorescence in situ hybridization signal patterns observed for genes and its significance in relation to the cutoff shown in Table 1.

Sample	Gene	1R2G (%)	1R1G (%)	Relation to cutoff value
1	CDKN2C	48.5	13.5	+
	CDKN2D	10.0	14.0	+
2	CDKN2C	6.0	9.5	-
	CDKN2D	8.0	10.0	-
3	CDKN2C	13.5	10.5	+
	CDKN2D	3.0	1.0	-
4	CDKN2C	11.5	12.0	+
	CDKN2D	1.0	3.0	-
5	CDKN2C	55.0	10.0	+
	CDKN2D	9.0	7.0	-

Table 3.

Percentages of CDKN2C loss in 4 medullary thyroid carcinoma samples according to research fluorescence in situ hybridization (FISH) and Clinical Laboratory Improvement Amendments (CLIA) FISH.

Sample	CDKN2C loss by Research FISH (%)	CDKN2C loss by CLIA FISH (%)
1	62	67
3	24	40
4	24	20
5	65	78

Discussion

Previous studies have shown that CDKN2C loss is associated with RET-mediated MTC [14–16]. Furthermore, the implication of CDKN2C and CDKN2D LOH in sporadic MTC was documented as well. Initial comparative genomic hybridization data indicated that 40% of MTC tumors have CDKN2C loss. Screening for CDKN2C loss with other methodologies is not feasible; therefore, we surmised that FISH would be the best way to screen for LOH and that this screening could be implemented into the clinical care of patients. In this study, we used a dual-color FISH assay to analyze chromosome 1p32 for CDKN2C loss and 19p13 for CDKN2D loss.

Given the nature of the FFPE tissue sections used in the present study, we expected to encounter some challenges when performing FISH. The MTC sections were heavily fibrous and had protein cross-linking due to formalin fixation, and these characteristics can mask DNA sites, which can affect the probes’ affinity to their binding sites. To account for these factors, the samples were rigorously pretreated at 95–98°C and then subjected to enzymatic digestion at 37°C, which maintained the integrity of the cells. In addition, we evaluated each batch of probes with high stringency. The validation eliminated any chance of false specific hybridization due to the probe design in addition to confirming that the probes bound to chromosome 1 and 19 proper regions with adequate fluorescence intensities. The protocol we used also proved to be highly efficient for hybridization with direct fluorochrome-labeled probes on FFPE tissue sections.

In conclusion, although various techniques are available to assess CDKN2C loss, only FISH fulfills the criteria as a fast and reliable method for doing so, especially for FFPE MTC samples. Of the 5 MTC samples we tested, 4 had CDKN2C loss; however, only 1 had loss of both CDKN2C and CDKN2D. We thus found frequent loss of CDKN2C in 80% of the MTC samples. Furthermore, the results of a cytogenetic CLIA test for CDKN2C were 100% concordant with our research FISH data. This study suggests that FISH offers a great advantage as a routine diagnostic tool for screening MTC patients.