9p21.3 Microdeletion involving CDKN2A/2B in a young patient with multiple primary cancers and review of the literature.

Germline pathogenic variants in CDKN2A predispose to various cancers, including melanoma, pancreatic cancer, and neural system tumors, whereas CDKN2B variants are associated with renal cell carcinoma. A few case reports have described heterozygous germline deletions spanning both CDKN2A and CDKN2B associated with a cancer predisposition syndrome (CPS) that constitutes a risk of cancer beyond those associated with haploinsufficiency of each gene individually, indicating an additive effect or a contiguous gene deletion syndrome. We report a young woman with a de novo germline 9p21 microdeletion involving the CDKN2A/CDKN2B genes, who developed six primary cancers since childhood, including a very rare extraskeletal osteosarcoma (eOS) at the age of 8. To our knowledge this is the first report of eOS in a patient with CDKN2A/CDKN2B deletion.

INTRODUCTION

CDKN2A and CDKN2B, both located on Chromosome 9p21, are involved in cell cycle regulation. CDKN2A encodes the tumor suppressors p16INK4A and p14ARF that regulate the cell cycle through the pRB and p53 pathways, respectively. CDKN2B encodes the protein p15INK4B, which, along with p16INK4A, acts as a cyclin-dependent kinase inhibitor and has been implicated in renal cell carcinoma predisposition in a single study (Jafri et al. 2015).

Pathogenic germline variants in CDKN2A are mainly associated with a risk of cutaneous melanoma that is more than 10-fold higher than the general population. Furthermore, CDKN2A variants affecting p16INK4A increase pancreatic cancer risk. Germline alterations in CDKN2A are also occasionally associated with a rare melanoma-astrocytoma syndrome, which predisposes to malignant melanomas and neural system tumors, including astrocytomas and meningiomas (Chan et al. 2017).

Because of their rarity, the significance of germline deletions involving both CDKN2A and CDKN2B is poorly described. Case reports of families or individuals with large deletions involving both CDKN2A and CKDN2B (Table 1) have reported neurofibromas, giant cell tumors of bone and multiple primary cancers including sarcomas, such as a malignant peripheral nerve sheath tumor (Baker et al. 2016; Chan et al. 2016), thus clinically mimicking phenotypes of cancer predisposition syndromes such as neurofibromatosis type 1 (NF1) or Li–Fraumeni syndrome (LFS). We report a case of a 21-yr-old woman without significant family history of cancer, who developed six primary malignant tumors (extraskeletal osteosarcoma, lymphoepithelial carcinoma, low-grade myofibroblastic sarcoma, melanoma, anaplastic pleomorphic xanthoastrocytoma, and high-grade astrocytoma with piloid features) and 16 benign or premalignant tumors (Tables 2 and 3). Single-nucleotide polymorphism (SNP) array and whole-genome sequencing (WGS) identified a de novo germline microdeletion involving the entire CDKN2A and CDKN2B gene regions, previously published in an overview of childhood cancer cases (Østrup et al. 2018).

Table 1.

Previous reports of CDKN2A–CDKN2B gene microdeletions

Case	Age at first malignant tumor	Clinical manifestations	Genomic analysis	Germline deletion	Family history
Petty et al. 1993	Not specified	By age 34: 8 melanomas, multiple atypical nevi, plexiform neurofibroma	FISH, Southern Blot, PFGE, PCR of STR polymorphisms	Unbalanced de novo cytogenic rearrangement of Chromosomes 5p and 9p, reported spanning at least 6 Mb of DNA; encompassing 2–3 Mb of Chromosome 9p
Frigerio et al. 2014	10	Neurothekeoma,anaplastic astrocytoma,tectal mesencephalic lesion,10 melanomas,neurofibroma	Microsatellite analysis, MLPA, aCGH	2.1 Mb (de novo),involving MTAP, IFN gene cluster, FOCAD, and partly CDKN2B-AS1;Seq(GRCh37) del(9)(p21.3) NC_00009.11:g.19934141-22069982del	Untested: Monozygotic twin sister with dysplastic nevi and slow growing cerebral lesion
Our case	8	Can be viewed in Tables 2 and 3	WGS, SNP array, MLPA	1.7 Mb (de novo);Seq(GRCh37) del(9)(p21.3) NC_00009.11:g.20657836-22353823del
Baker et al. 2016	18 (proband's son)	Pontomedullary primitive neuroectodermal tumor	Whole-genome oligonucleotide array CGH, SNP array	At least 1.5 Mb,involving MTAP and partly FOCAD;Seq(GRCh37) del(9)(p21.3) NC_00009.11:g.20951885-22447709del;further, two CNVs of unknown significance and an amplification of unknown significance	Carriers (proband and proband's sister): neurofibromas, squamous papilloma, benign histiocytoma, melanoma, atypical nevi;eight untested family members: leukemia, chondrosarcoma, melanoma, cervical cancer, four members with unspecified cancer;noncarrier: melanomas, atypical nevi, lipoma
Chan et al. 2017	31	Pleomorphic xanthoastrocytoma,2 diffuse astrocytomas, peripheral nerve sheath tumor	Targeted NGS of 479 cancer-associated genes, including selected introns	1.1 Mb, partly involving MTAP;Seq(GRCh38) del(9)(p21.3) NC_00009.11:g.21700000-22800000del	Untested family members: oral cancer (father, paternal uncle),glioblastoma (sister),melanoma (paternal uncle, grandfather),brain cancer (paternal grandfather)
Pasmant et al. 2007; Bahuau et al. 1998	Not specified	Neural system tumors,melanomas	Bahuau et al.: heterozygosity mapping based on microsatellite markersPasmant et al.: STS real-time PCR-based gene dose mapping, long-range PCR and nucleotide sequencing	403 kb;approximate coordinates: 9:21,926,315-22,329,545	13 confirmed carriers by Bahuau et al. (3 of which were tested by Pasmant et al.): a total of 8 melanomas, at least 6 neural system tumors, and 1 unspecified neoplasm
Chan et al. 2016	38	Laryngeal squamous cell carcinoma,high-grade malignant peripheral nerve sheath tumor	WGS, validated by qPCR	∼270 kb;the deletion is partially truncating the flanking MTAP and CDKN2B-AS	Untested family members: giant cell tumor of bone (brother), possibly colon cancer (mother), possibly liver cancer (father)

(FISH) Fluorescence in situ hybridization, (PFGE) Pulsed-field gel electrophoresis, (PCR) polymerase chain reaction, (STR) short tandem repeat, (MLPA) multiplex ligation-dependent probe amplification, (aCGH) array comparative genomic hybridization, (WGS) whole-genome sequencing, (SNP) single-nucleotide polymorphism, (CNV) copy number variant, (NGS) next-generation sequencing, (STS) sequence-tagged sites, (qPCR) quantitative PCR.

Table 2.

Clinical manifestations and treatment of malignant and premalignant tumors in our proband

Age (y)	8	13	15	17	18			19	20	22
Diagnosis	Extraskeletal osteosarcomaa	Parotid metastasis of lymphoepithelial carcinoma	Anaplastic pleomorphic xanthoastrocytoma	Low-grade myofibroblastic sarcomaa	Relapse of APXA	Melanoma with micrometastasis in one axillary lymph node	High-grade astrocytoma with piloid features	Premalignant atypical neurofibroma (several tumors)	Premalignant atypical neurofibroma	Premalignant atypical neurofibromaa
Location	Right upper arm	Next to right ear	Right temporal lobe	Right gastrocnemius muscle	Right temporal lobe	Left shoulder	Left side of hypothalamus	Left brachial plexus	Left side of neck	Right neuroforamina, Th2-3
Symptoms	Swollen tumor	Swollen, painless lymph node	Headache, emesis, paresthesia in right parietal region	Asymptomatic	Asymptomatic	Change of color, pain	Asymptomatic	Swollen tumor	Swollen tumor	Asymptomatic
Preoperative imaging	Ultrasound, X-ray, MRI, PET/CT	Ultrasound, PET/CT	MRI, PET/CT	PET/CT, MRI, ultrasound	MRI	None	MRI, PET/CT	MRI	MRI	MRI
Preoperative biopsy	Yes	No	No	No	No	No	Yes	No	No	Biopsied ∼2 yr before resection
Treatment	Preoperative chemotherapy, resection, postoperative chemotherapy	Resection, postoperative radiation therapy and chemotherapy	Resection	Resection	Resection	Resection, reresection, sentinel node with resection of 2 nodes	Radiation therapy, chemotherapy	Resection	Resection	Resection

(APXA) Pleomorphic xanthoastrocytoma, (MRI) magnetic resonance imaging, (PET/CT) positron emission tomography/computed tomography.

aHistopathological descriptions

Extraskeletal osteosarcoma (eOS): A mesenchymal tumor with no connection to the skeletal system consisting of very pleomorphic cells and osteoclast like giant cells. Without immunohistochemical features of specific differentiation. In the pretreatment biopsy there was identification of osteoid. In the posttreatment resection specimen, there was treatment response but no osteoid could be identified. The diagnosis of eOS therefore rests on the pretreatment biopsy, and the diagnosis of an undifferentiated pleomorphic sarcoma cannot be excluded.

Low-grade myofibroblastic sarcoma: A mesenchymal tumor consisting of infiltrative, cellular fascicles of myofibroblastic spindle cells with focal nuclear atypia. Immunohistochemistry with smooth muscle actin (SMA) and calponin positivity. Desmin was negative.

Atypical neurofibromatous neoplasms of uncertain biological potential (ANNUBP): A wavy spindle cell neoplasm with a variably myxoid to collagenous stroma (shredded carrots). There was some cytological atypia, areas with hypercellularity and mitosis (but no more than 2 mitosis/10 high-power field [HPF]). Immunohistochemistry was positive for S-100.

Table 3.

Overview of benign tumors in our proband, including type of tumor, age, and location

Tumor type	Number of tumors	Age at diagnosis	Location
Neurofibromas	3	14	Pelvic region, right femoral nerve (resected)
		17	Left brachial plexus (resected)
			Peripheral nerve in right forearm (proximal and ulnar located) (resected)
Aneurismatic bone cyst	1	19	Right middle cranial fossa near the cavernous sinus (embolized)
Cellular neurothekeoma	1	20	Left side of neck (resected)
Benign soft tissue tumor	1	20	Left side of neck (resected)
Tumors identified radiologically without biopsy or resection	9	18	Intermuscular process in relation to right schiatic nerve near femur midshafta
		19	In relation to left femoral nerve/femoral condyleb
		20	Left cerebellopontine angle in relation to trigeminal nerve
			Process posteriorly to right m. psoasb
			Anteriorly to L5/S1, right sideb
			In the right neuroforamina L4/L5b
		21	Left neuroforamina C6/C7b
			Intraspinal process, probably intradural and extramedullary, posterior to C7/Th1 discc
			Left to sternum

aTentative diagnosis: neurofibroma.

bTentative diagnosis: schwannoma.

cTentative diagnosis: meningioma.

RESULTS

Clinical Presentation and Family History

At age 8 the proband, a previously healthy, nondysmorphic female with normal development, Iraqi heritage, and consanguineous parents, developed a swelling in her right upper arm. The tumor was biopsied, and histology, immunohistochemistry, and electron microscopy identified the tumor most likely to be an extraskeletal osteosarcoma (for histopathological description, view Table 2 footnotes). Treatment consisted of preoperative chemotherapy initially following the EpSSG NRSTS 2005 protocol (fosfamide, Adriamycin), without response, therefore altered with response to EURAMOS protocol (cisplatin, doxorubicin, and high-dose methotrexate) followed by resection and postoperative chemotherapy.

At age 13, she developed a swollen, painless lymph node next to the left ear. Resection of the lymph node showed a metastasis of an Epstein–Barr virus (EBV)-negative lymphoepithelial carcinoma. Tonsillectomy, adenoidectomy, and biopsies from nasopharynx, tongue root, and pharyngoepiglottic fold were unable to confirm the site of the primary tumor; however, the rhinopharynx was considered the most likely origin. A few months later, a relapse was identified adjacent to a resected lymph node and treated with radiation therapy of nasopharynx and the neck bilaterally (34 × 2 Gy, 6 F/W) and concomitant chemotherapy (cisplatin). In the diagnostic process of the lymphoepithelial carcinoma, a positron emission tomography/computes tomography (PET/CT) scan identified a process located in the right pelvis near the iliacus muscle. A 12-mo treatment follow-up scan showed slow growth. The pathology of an ultrasound-guided biopsy identified the process as a neurofibroma, which was completely and uneventfully resected. Pathological examination including immunohistochemistry of biopsy and resected tumor showed a tumor consisting of a single nodulus containing small spindle-shaped cells with regular oval nuclei and a normal staining pattern, showing no atypia, necrosis, or mitoses. Somatic genetic testing has not been performed, but pathology showed a classical pattern of neurofibroma. No café-au-lait spots or freckling were present, but the patient had multiple regular skin nevi throughout her body, including a Spitz nevus which was subsequently removed.

At age 15, the proband experienced severe headache, intermittent emesis, and paresthesia in the right parietal region. A brain magnetic resonance imaging (MRI) scan showed a large tumor in the right temporal lobe measuring ∼25 × 17 mm. The tumor was surgically removed in toto and identified as an anaplastic pleomorphic xanthoastrocytoma (APXA), which was confirmed by genome-wide methylation analysis using the Heidelberg classifier (molecularneuropathology.org). The tumor showed a somatic BRAF mutation (p.Val600Glu) and loss of heterozygosity for the CDKN2A/CDKN2B gene region. Exome sequencing was performed, and all tumor-specific variants can be viewed in Table 4. No further treatment was given because of her, at this time, extensive chemotherapy and radiation exposure, including consideration of future treatment options, as the tumor was BRAF mutated. Remission lasted a year before a local relapse occurred. Once again, a complete surgical resection was performed with no further therapy added. In the same year, at age 17, a surveillance PET/CT scan showed increased FDG uptake in the right gastrocnemius muscle, reflecting a low-grade myofibroblastic sarcoma. Histopathological description can be viewed in the Table 2 footnotes, and histopathological images in Supplemental Figure S1. The tumor was surgically removed with clean resection margins. No further therapy was given.

Table 4.

Overview of germline deletion and somatic variants in our proband

Germline/somatic (tumor)	Gene(s)	Chromosome	HGVS DNA reference	HGVS protein reference	Variant type	Predicted effect	dbSNP/dbVar ID	Genotype
Germline	a	9p21.3	seq(GRCh37) del(9)(p21.3) NC_00009.11:g.20657836-22353823del	N/A	Deletion	Pathogenic	N/A	Heterozygous
Somatic (APXA)	BRAF	7q34	NM_004333.4: c.1799T > A	p.Val600Glu	Missense	Pathogenic	N/A	N/A
Somatic (APXA)	SP140	2q37.1	NM_007237.4: c.1994G > C	p.Arg665Thr	Missense	N/A	N/A	N/A
Somatic (APXA)	ALDH5A1	6p22.3	NM_001080.3: c.838_839delAT	p.Ile280fsa37	Frameshift	N/A	N/A	N/A
Somatic (APXA)	SCUBE2	11p15.4	NM_020974.2: c.1558G > A	p.Val520Ile	Missense	N/A	N/A	N/A
Somatic (APXA)	GRP	18q21.32	NM_002091.3: c.239C > T	p.Ala80Val	Missense	N/A	N/A	N/A
Somatic (APXA)	WDR62	19q13.12	NM_173636.4: c.1917_1918delCCinsGT	p.Gln640a	In-frame	N/A	N/A	N/A
Somatic (high-grade astrocytoma with piloid features)	NF1	17q11.2	NM_000267.3:c.6789_6792del	p.Tyr2264Thrfsa5	Frameshift	Pathogenic	dbSNP: rs863224836	N/A
Somatic (high-grade astrocytoma with piloid features)	NF1	17q11.2	NM_000267.3:c.1393-1G > A	p.?	Splice acceptor	Likely pathogenic	dbSNP: rs1131691131	N/A
Somatic (latest ANNUBP)	CDKN2A	9p21.3	c.191_212delinsCGTGG	p.Leu64fsa50	Frameshift	N/A	N/A	N/A
Somatic (latest ANNUBP)	NF1	17q11.2	c.1026_1035del	p.Ile342fsa31	Frameshift	N/A	N/A	N/A

Somatic sequence data of biopsy of high-grade astrocytoma with piloid features:

NGS analysis was performed using a neuropanel covering hotspots in the following genes: BRAF, FGFR1, H3F3A, Hist1H3B, Hist1H3C, IDH1, IDH2, PIK3CA, PIK3R1, MET, NRAS, SMO, and TERT promoter. The panel covers the whole coding sequence of the following genes: ATRX, CDKN2A, CDKN2B, CDKN2C, CIC, EGFR, FUBP1, NF1, NF2, NOTCH1, PTEN, RB1, and TP53. Further, there was an examination of codeletion of 1p and 19q. Sensitivity of mutations in the analysis is 5% tumor cell nuclei.

Somatic sequence data of resection of the latest atypical neurofibromatous neoplasm of uncertain biological potential (ANNUBP):

Next-generation sequencing mutation analysis and copy-number variation analysis was performed using a neuropanel covering the following genes: ATRX^, BRAF*, CDKN2A“^, CDKN2B"^, CDKN2C^, CIC^, EGFR”^, FGFR1*, FUBP1^, H3F3A*, Hist1H3B*, Hist1H3C*, IDH1*, IDH2*, MET*, NF1^, NF2^, NOTCH1^, NRAS*, PIK3CA*, PIK3R1*, PTEN^, RB1^, SMO*, TERT promoter*, and TP53^. Further, there was an examination of 1p monosomy and codeletion of 1p and 19q. Sensitivity of mutations in the analysis is 5% tumor cell nuclei. (*) The analysis covers most frequent hotspot mutations, (“) the analysis covers copy-number variants, (^) the analysis covers the whole coding sequence.

aCDKN2A, CDKN2B, CDKN2A-AS1, CDKN2B-AS1, ERVFRD-3, FOCAD, FOCAD-AS1, HACD4, IFNA1, IFNA10, IFNA13, IFNA14, IFNA16, IFNA17, IFNA2, IFNA21, IFNA22P, IFNA4, IFNA5, IFNA6, IFNA7, IFNA8, IFNB1, IFNE, IFNW1, IFNWP15, IFNWP18, IFNWP2, IFNWP5, IFNWP9, KHSRPP1, KLHL9, MIR31, MIR31HG, MIR491, MTAP, TUBB8P1, UBA52P6.

At age 18, she was diagnosed with her fifth primary cancer, a superficial spreading malignant melanoma on her left shoulder with a tumor thickness of 1.3 mm. Sentinel node examination showed micrometastasis in one of two sentinel lymph nodes in the axillary region, and surgical resection was performed. The patient has Fitzpatrick skin phototype 4 and no record of sunburns in childhood. Further, there are no other known melanoma cases in her family.

In the same year, a brain surveillance MRI revealed an asymptomatic hypothalamic brain tumor. Following initial observation, progression of tumor size led to a biopsy being performed. Based on histology and genome-wide methylation analysis the tumor was classified as a high-grade astrocytoma with piloid features. Like the APXA, loss of heterozygosity for the CDKN2A/CDKN2B gene region was demonstrated. The high-grade astrocytoma with piloid features further showed a pathogenic and a likely pathogenic somatic mutation in the NF1 gene (NM_000267.c.6789_6792del and NM_000267.c.1393-1G > A, respectively) and no somatic BRAF mutation was identified, supporting the diagnosis of a new primary tumor. Information regarding somatic analysis by next-generation sequencing (NGS) can be viewed in the Table 4 footnotes. Treatment of the second astrocytoma consisting of radiation (1.8 Gy × 30 F/54 Gy) and chemotherapy (temozolomide) was initiated. Evaluation after radiotherapy and two series temozolomide showed pseudoprogression of the tumor. Early follow-up scans (11 mo) show a stable tumor size (Fig. 1).

Figure 1.

Latest magnetic resonance imaging of the high-grade astrocytoma with piloid features, showing stable tumor size.

In addition to the six malignant tumors described above, the proband has had three premalignant atypical neurofibromas, all surgically resected, and 13 benign tumors since age 14. The benign tumors include at least three neurofibromas (none plexiform), all of which resected, and five tumors thought to be schwannomas based on MRI. The latest atypical neurofibromatous neoplasm of uncertain biological potential (ANNUBP) was examined by NGS analysis (for details regarding NGS analysis, view the Table 4 footnote), showing heterozygote deletion of CDKN2A/B and small indels in both NF1 and CDKN2A causing truncating variants (for specific variants, view Table 4). An overview of malignant, premalignant, and benign tumors is presented in Tables 2 and 3.

At present, the proband is receiving symptomatic treatment and frequent imaging of the high-grade astrocytoma with piloid features and the multiple benign peripheral nerve sheath tumors. In addition, a surveillance program comprising of quarterly MRI of the neuroaxis and annually melanoma screening has been followed since the age of 16 yr.

Genomic Analyses

At age 15, after diagnosis of the anaplastic pleomorphic xanthoastrocytoma, genomic analysis was performed. Exome sequencing did not show any pathogenic variants. A CytoScan HD SNP array found a heterozygous germline deletion on Chromosome 9p21.3, including CDKN2A, CDKN2B, and the IFN gene family. Multiplex ligation-dependent probe amplification (MLPA) analysis and parental testing confirmed a de novo CDKN2A/CDKN2B deletion (Østrup et al. 2018).

At age 19, the proband was referred to genetic counseling and offered WGS as part of a research project, STAGING (Byrjalsen et al. 2020). A written consent for publication was obtained in the STAGING project. WGS mapped the deletion and determined the size to be 1.7 MB (seq(GRCh37) del(9)(p21.3) NC_00009.11:g.20657836-22353823del), further including the genes FOCAD, MTAP, and CDKN2B-AS1 (Table 4; Fig. 2). For determination of the exact breakpoints, fine mapping of the deletion break points is necessary. No additional known pathogenic variants in genes suspected of cancer predisposition syndrome (CPS) were identified.

Figure 2.

Whole-genome sequencing demonstrating the heterozygous germline deletion on Chromosome 9p21.3 containing CDKN2A and CDKN2B.

DISCUSSION

In a patient with a germline microdeletion and a severe cancer predisposition, difficulties arise regarding evaluating the contribution of the individual genes, and the manifestations could be suggestive of an additive genetic effect or a contiguous gene deletion syndrome, meaning that tumor proneness possibly could relate to the size of the deletion.

In addition to CDKN2A and CDKN2B, the deletion in our proband involved the genes FOCAD, CDKN2B-AS1, MTAP, and the IFN gene cluster. However, these genes do not yet have a well-described role in terms of cancer predisposition. FOCAD encodes the tumor suppressor focadhesin, and one study has found germline deletions in the FOCAD gene to be associated to colorectal cancer (Weren et al. 2015). CDKN2B-AS1, previously designated ANRIL, is an antisense noncoding RNA partially located within the CDKN2A–CDKN2B locus suspected of suppressing the expression of CDKN2A and CDKN2B (Visel et al. 2010; Yap et al. 2010; Kotake et al. 2011). In a family-based association study by Pasmant et al., a SNP rs2151280 within CDKN2B-AS1 was found to be associated with the number of plexiform neurofibromas (PNFs) in NF1 patients (Pasmant et al. 2011). However, further studies are necessary before a significance of CDKN2B-AS1 deletion in terms of tumor predisposition can be established.

To our knowledge, neither the IFN gene cluster nor MTAP is known to cause cancer predisposition. However, Linsley et al. finds that homozygous deletion of the IFN gene cluster may contribute to reduced tumor destruction by the immune system (Linsley et al. 2014), thus potentially leading to a worse prognosis. Exploring the role of MTAP, a gene encoding a key enzyme in the adenine and methionine salvage pathway, two reports have shown that cancer cells with homozygously deleted MTAP can be sensitive to PRMT5 inhibition, which may be at potential therapeutic strategy (Kryukov et al. 2016; Mavrakis et al. 2016). Further, a report by Camacho-Vanegas et al. (2012) found that germline splice variants of MTAP can result in a distinctive clinical phenotype with diaphyseal medullary stenosis and increased risk of malignant fibrous histiocytoma, and has suggested that MTAP function as a tumor suppressor.

A total of six families describing germline deletion of the entire CDKN2A/CDKN2B locus have previously been described, comprising 20 confirmed carriers with a total of at least 32 malignant or premalignant cancers and 13 benign tumors. The cases are presented in Table 1 according to the size of the deletion. A large span is seen in age at presentation and severity of cancer manifestation in these cases, possibly reflecting differences in deletion size and thus supportive of the contiguous gene deletion syndrome. Regarding co-deleted genes, the report by Frigerio et al. (2014) describing a girl with a severe tumor manifestation and the largest well-determined deletion (2.1 Mb), involves FOCAD, MTAP, the IFN gene cluster, and as well partly CDKN2B-AS1. The case by Baker et al. (2016) describes a deletion involving MTAP, and the approximate coordinates of the deletion suggest affected gene function of FOCAD as well. Similarly, the approximate coordinates of the deletion in the case by Chan et al. (2017) suggests deletion of MTAP (Chan et al. 2017).

The analysis of a potential contiguous gene deletion syndrome is limited by the small number of cases, by available clinical data, and by the fact that only one previous study (Chan et al. 2016) has performed extensive sequencing by WGS and thereby excluded alterations in additional cancer predisposition genes. Additional cancer predisposition alterations may be present in the case by Petty et al. (1993) as the CDKN2A/CDKN2B deletion was the result of an unbalanced translocation between Chromosome 5p and 9p. Further studies will be necessary to evaluate the possibility of a contiguous gene deletion syndrome.

Regarding somatic NF1 variants, the high-grade astrocytoma with piloid features had two somatic NF1 variants, as previously mentioned, none of which identified germline including screening for mosaicism. Analysis of the APXA with whole-exome sequencing (WES) did not show NF1 alterations. None of the additional tumors has been analyzed. An increased propensity for somatic NF1 alterations when a germline CDKN2A (p14ARF) alteration is present, has previously been suggested (Rhodes et al. 2019), but further investigations have to be performed to confirm this interaction. However, both gliomas and one of the atypical neurofibromas have shown homozygous loss of CDKN2A/CDKN2B, which suggests CDKN2A/CDKN2B as driving for tumor development in the neural system tumors as well as the patient's tumors not typical for NF1.

Potential consequences of cancer treatment must be discussed as well. In our case, three of the cancers have been treated with alkylating chemotherapeutics, beginning in the treatment of the proband's very first cancer at age 8 yr. Alkylating chemotherapeutics are known to induce alterations in DNA, and similarly radiation therapy can increase risk of cancer in the radiation field. This patient has received radiation therapy to the neck and rhinopharynx in the treatment of the lymphoepithelial carcinoma (2 Gy × 34) and in treatment of the latest tumor, the high-grade astrocytoma with piloid features (54 Gy). Thus, chemotherapy and radiation therapy could also have contributed to subsequent tumor development.

Multiple pediatric cancer predisposition syndromes have recommended surveillance protocols (Frebourg et al. 2020; Durno et al. 2021). Interestingly, two of the six malignant tumors (the low-grade myofibroblastic sarcoma and the high-grade astrocytoma with piloid features) and the relapse of APXA in the patient presented were identified asymptomatically. Early asymptomatic diagnosis may be vitally important for patients at high risk of multiple cancers. When purely surgical cures are possible it ensures not just decreased mortality and morbidity, but also that dosage-limited therapies such as radiation and less toxic chemotherapy are still options for subsequent, harder-to-treat cancers. In addition, surveillance can prevent malignancy by surgical removal of premalignant tumors.

Difficulties arise in establishing recommendations regarding surveillance for patients with CDKN2A–CDKN2B microdeletions, as we are limited by the rarity of this syndrome and the large span in age, cancer type, and severity between the cases. In our case, the patient fulfills the Chompret criteria for LFS. Thus, a surveillance protocol resembling LFS protocols, such as the Toronto protocol (Villani et al. 2016), should be considered. However, it should be noted that LFS surveillance is recommended for confirmed germline TP53 variant carriers. Although optimal strategies remain unknown, we encourage evaluation of each patient's clinical presentation, family history, and the deletion when planning a surveillance protocol.

In summary, this case illustrates that pediatric cancer patients with a phenotype resembling LFS should be offered additional extensive genetic evaluation if no germline TP53 variant is identified, to exclude structural alterations or variants in other CPS genes. Similarly, as most of the presented cases of CDKN2A/CDKN2B germline deletions have reported nerve sheath tumors and/or gliomas, including our own case, the CDKN2A/CDKN2B locus should also be assessed in patients with neural system tumors without pathogenic variants in NF1, NF2, LZTR1, and SMARCB1.

By identifying such pathogenic variants, affected patients will be available for targeted surveillance regimens, which potentially can identify any additional tumors at an early and treatable stage. Further, identification permits presymptomatic testing of relatives at risk, as well as prenatal diagnosis.

METHODS

Methods of WGS (and RNA sequencing) follow the STAGING project and can be viewed in detail in the study by Byrjalsen et al. (2020). Further, details of exome sequencing, the CytoScan HD SNP array, and MLPA analysis can be viewed in the study by Østrup et al. (2018).

The cancer panel–specific coverage in WGS is more than 20×: 98.53%

ADDITIONAL INFORMATION

Data Deposition and Access

Consent was not given for deposition of the data raw file of WGS data or of exome sequencing of APXA. Germline data are deposited in DECIPHER (https://www.deciphergenomics.org/) and can be found under accession number 480874.

Ethics Statement

Written patient consent was obtained. The project has been approved by the regional ethics committee (H-15016782).

Acknowledgments

We thank our patient for contributing to this study.

Author Contributions

K.W. provided genetic counseling and obtained informed consent. K.W. and M.R.J. wrote the manuscript. M.B. imaged the WGS data. M.R.J. provided the clinical image. All authors contributed to revision of the manuscript and approved the final manuscript and its submission to Cold Spring Harbor Molecular Case Studies.

Funding

This study was financially supported by the Independent Research Fund Denmark (M.R.J.; https://dff.dk/) and the European Union's Interregional Öresund-Kattegat-Skagerrak grant (https://interreg-oks.eu/). This work is part of the nationwide research program Childhood Oncology Network Targeting Research, Organization & Life expectancy (CONTROL) and supported by the Danish Cancer Society (R-257-A14720), the Danish Childhood Cancer Foundation (2019-5934 and 2020-5769), and the NEYE foundation. This work is part of Interregional Childhood Oncology Precision Medicine Exploration (iCOPE), a cross-Oresund collaboration between University Hospital Copenhagen, Rigshospitalet, Lund University, Region Skåne, and Technical University Denmark (DTU), supported by the European Regional Development Fund.

Competing Interest Statement

The authors have declared no competing interest.

Referees

D Gareth Evans

Maria-Isabel Achatz

Anonymous