Progress in Endocrine Neoplasia

Endocrine tumors are rare, most often occurring sporadically and in women. There are more differences than similarities among the tumors, not only concerning anatomic location and secretory function, but the frequency of hereditary disease (rare in adrenal cortical and differentiated thyroid tumors; uncommon in pituitary tumors; and common in neuroendocrine tumors of the adrenal, the pancreas, and the thyroid) and the frequency of malignant disease (virtually nonexistent in pituitary tumors; 10% to 30% in tumors of the adrenal cortex, adrenal medulla, and follicular thyroid; and 50% to 100% in neuroendocrine tumors of the pancreas and thyroid). Except for thyroid tumors, which are usually visible or palpable on physical examination, endocrine tumors are most often occult and become evident either because of a secretory malfunction, a space occupancy problem, or their incidental detection on imaging studies. The spectrum of endocrine tumors is almost completely expressed within the multiple endocrine neoplasia (MEN) syndromes, MEN1 and MEN2. MEN1 includes tumors primarily of the pituitary, parathyroids, and pancreatic islets and less often tumors of the thyroid follicular cells, adrenal cortical cells, and carcinoid cells. MEN2 includes tumors of the neuroendocrine thyroid, the parathyroids, and the adrenal medulla. The discovery of the genetic mutations causing the MEN1 and MEN2 syndromes has clarified much about the molecular biology of the syndromes and their component sporadic tumors, and, in many cases, has led to improved methods of diagnosis and treatment. As is often the case, information gained from the basic and clinical research of rare tumors serves as a guide to investigative strategies of more common neoplasms.

Surgical resection is the treatment for many patients with primary endocrine tumors, but molecular targeted therapeutics have replaced chemotherapy as frontline therapy for patients with advanced cancers of the pancreas and thyroid. The FDA has approved several molecular targeted therapeutics for certain cancers, based on phase III prospective, randomized, placebo-controlled clinical trials that showed significantly improved progression-free survival (PFS) in patients with metastatic disease (Table 1). Unfortunately, the drugs are associated with significant toxicities, and they are expensive. Most importantly, they have failed to demonstrate significant improvement in overall survival (OS). There is a critical need for new drugs with great specificity for oncogenic targets and an equally urgent need to understand the mechanisms of resistance to molecular targeted therapeutics—so far lacking for most solid organ malignancies—as this is key to the development of rescue therapy for tumors that become resistant to an initial therapeutic (1, 2).

The Cancer Genome Atlas (TCGA) studies of the NCI and the National Human Genome Research Institute have generated comprehensive, multidimensional maps of the key genomic changes in types of human cancer. The importance of these studies cannot be overemphasized, as they have expanded our knowledge of the molecular biology and pathogenesis of selected tumors, often defining a relationship between genotype and phenotype and yielding useful information regarding potential diagnostic tests and new therapeutics. Whether TCGA studies will lead to a broad array of curative molecular targeted therapeutics remains an open question. TCGA recently reported study results of adrenal cortical carcinoma (ACC), papillary thyroid cancer, and pheochromocytoma and paraganglioma. The complete study results are available in publications, either in print form or online, as videos (3–5).

ACC occurs in a bimodal pattern, developing during childhood or the fourth decade of life (6). The majority of adrenal cortical carcinomas are sporadic, but some occur in a hereditary pattern as a component of diseases, such as the Li–Fraumeni syndrome, the Lynch syndrome, and the Beckwith–Wiedemann syndrome (7–9).

In the recently published TCGA study of 91 patients with ACC, the female-to-male ratio was 1.8, and 57% of the tumors functioned, most often secreting cortisol (5). Median OS was 78 months, and 5-year survival was 59%; however, locally invasive and metastatic ACCs were associated with a 5-year survival of 22% (5). With the increased use of sensitive imaging studies came an increased incidental detection of occult adrenal tumors, the incidence increasing with age, being as high as 10% in the elderly (10). Complete surgical resection is the only cure for ACCs; however, when this is not possible, tumor debulking may relieve symptoms caused by hormonal excess. There is a direct correlation between the size of an adrenal tumor and the presence of malignancy; approximately 80% of tumors larger than 6 cm in diameter are malignant (11). Most surgeons limit laparoscopic resection of adrenal tumors to those less than 6 cm in size and with no evidence of malignancy. In this CCR Focus, Payabyab and colleagues stress the importance of avoiding laparoscopic biopsy, or resection, of potential or proven ACC, citing the likelihood of peritoneal seeding, a fatal complication (12).

Mitotane is the only drug approved by the FDA for treatment of advanced ACC; however, prospective trials of mitotane alone in patients with advanced disease showed partial responses of 5% to 30%, with no survival benefit (13). Furthermore, the substantial toxicity of mitotane has limited its use. There have been two multicenter retrospective trials of adjuvant mitotane in patients with ACC. One trial of 177 patients suggested a prolonged recurrence-free survival; however, the second trial of 207 patients showed no improved PFS or OS (13–15). Phase II trials of combination chemotherapy, with or without mitotane, showed minimal benefit in patients with advanced or metastatic disease (16–18). In a prospective phase III trial, 304 patients with advanced ACC were randomized to receive either a combination of etoposide, doxorubicin, and cisplatin or mitotane plus streptozotocin. The response rate (23.2% vs. 9.2%; P < 0.001) and the PFS (5 months versus 2.1 months; P < 0.001) were higher in patients receiving the three-drug combination, compared with those receiving mitotane and streptozotocin; however, there was no improvement in OS (19).

The mitotic rate of ACC cells, as is the case for most endocrine and nonendocrine tumor cells, is directly proportional to the clinical behavior of the parent tumor (20–22). Weiss and colleagues studied the relationship between histologic parameters and survival in 41 patients with ACC. Only mitotic rate was significantly associated with survival. Accordingly, ACCs were designated as high grade or low grade based on whether they had greater than or less than 20 mitoses per 50 high-powered fields (20). TCGA study of ACC confirmed and expanded the Weiss and colleagues' classification by integrating tumor subsets identified across the DNA copy number and mutations, DNA methylation, mRNA expression, and miRNA expression platforms (5). In a cluster of cluster (CoC) analysis, they designated three tumor subtypes, CoCI, CoCII, and CoCIII, with respective disease progression rates of 7%, 56%, and 97%. At the time of the study, the median event-free survival was not yet reached for CoCI, although it was 38 months for CoCII, and 8 months for CoCIII. Loss of large segments of DNA followed by whole-genome doubling was also a marker of tumor progression (5).

Chemotherapy is currently the frontline treatment for patients with advanced ACC, and much effort is being devoted to the creation of effective molecular targeted therapies and immunotherapies. It is hoped that the trove of molecular data generated by TCGA study of ACC, and by other investigators in the field, will lead to the development of curative drugs.

Pheochromocytomas and paragangliomas, referred to collectively as pheochromocytomas, are chromaffin-rich tumors arising from neural crest–derived sympathetic lineage cells in the adrenal medulla or from paraganglia in the thorax, abdomen, or pelvis (23). Sympathetic-derived tumors secrete epinephrine and norepinephrine, whereas parasympathetic-derived tumors, arising in the head and neck along the vagus or glossopharyngeal nerves, do not (23, 24). The annual incidence of pheochromocytomas is approximately two cases per 1 million individuals, and the tumor occurs in 0.6% of patients with hypertension and in 5% of patients with incidentally discovered adrenal masses (25, 26). Pheochromocytomas are most often sporadic, but an increasing number is reported to be hereditary. The associated age of onset of pheochromocytoma is 43.9 years in sporadic cases and 24.9 years in hereditary cases (27).

The diagnosis of hormonally active pheochromocytomas depends on the detection of increased plasma levels of free metanephrines and methoxytyramine (28, 29). Surgical resection, under careful pharmacologic preparation, is the treatment of pheochromocytomas; the procedure is performed either by laparoscopic resection or open resection. Depending on the anatomic location, from 10% to 40% of pheochromocytomas are malignant, and the associated 5-year survival is 50% (30). Treatment options for locally advanced or malignant pheochromocytomas include chemotherapy, radiotherapy, ¹³¹I-MIBG, and molecular targeted therapeutics. The regimen of cyclophosphamide, vincristine, and dacarbazine is the frontline therapy used the most often for patients with advanced disease, even though there has been no prospective clinical trial evaluating its efficacy (31). In a retrospective evaluation of temozolomide in 15 patients with metastatic pheochromocytoma, the PFS was 13.3 months; 5 patients had partial responses, 7 patients had stable disease, and 3 patients had tumor progression. Interestingly, the partial responses occurred only in patients with SDHB mutations (32). In a retrospective study of sunitinib in 17 patients with malignant pheochromocytomas, 3 patients had a partial response and 8 patients had stable disease. Most patients who had a clinical benefit carried the SDHB mutation (33).

A recent TCGA study of the genomic landscape of 173 pheochromocytomas found that considering somatic mutations, DNA chromosome breaks, mRNA proliferation, and methylation, pheochromocytomas have a relatively quiet genome; however, they have the highest degree of heritability of any human tumor (34). In 12 familial syndromes, linked to a specific genetic locus, there is an increased risk of developing pheochromocytomas (29). In 65% of pheochromocytomas, mutations have been discovered in 19 mutually exclusive susceptibility genes. The mutations are either germline (SDHA, SDHB, SDHC, SDHD, RET, VHL, NF1, MAX, EGLN1, and TMEM127) or somatic (HRAS, EPAS1, ATRX, CSDE1, GPR128, SETD2, ARNT, FGFR1, and BRAF). RET, VHL, and NFI occur as both germline mutations and somatic mutations (23, 34, 35).

Pheochromocytomas can be divided into three clusters, depending on whether they have mutations in genes that alter proteins constituting the Krebs cycle (SDHA, SDHB, SDHC, SDHD, SDHAF2, MDH2, and FH), in genes associated with the hypoxic response (VHL and EPAS1), or in genes linked to signaling in the RAS/RAF, MAPK, or mTOR pathways. The mechanisms of oncogenesis remain unknown for a large number of pheochromocytomas.

Pheochromocytoma was the first human tumor shown to be caused by a germline mutation of SDHD, a gene that encodes a metabolic enzyme, and also the first human tumor shown to have activating mutations in HIF2A (36, 37). In this CCR Focus, Jochmanova and Pacak focus on the metabolic aspect of pheochromocytomas, describing them as the first metabolic endocrine tumors and discussing how metabolic alterations generated by genetic mutations suggest avenues for personalized cancer management (38).

Thyroid cancer is the most common endocrine malignancy, and more than 95% of the tumors arise from follicular cells, most often as differentiated thyroid carcinomas, including papillary (85%) and follicular thyroid carcinomas (2%–5%), and less often as poorly differentiated (1%–3%) or anaplastic thyroid carcinomas (1%–3%). Generally, papillary thyroid carcinomas and follicular thyroid carcinomas are indolent clinically, whereas poorly differentiated and anaplastic thyroid carcinomas are highly aggressive with mean survivals of 3.2 years and 6 months. The incidence of thyroid cancer has tripled over the last three decades, primarily due to the increased detection of small papillary thyroid carcinomas on imaging studies (39, 40). Papillary thyroid carcinomas less than 1 cm in size (papillary microcarcinomas) occur in up to 30% of adults. The majority of microcarcinomas are stable over time and can be managed expectantly by monitoring tumor growth (41, 42). Approximately 5% of thyroid cancers are derived from the neural crest as medullary thyroid carcinoma.

The encapsulated variant of the follicular variant of papillary thyroid carcinoma accounts for 10% to 20% of all thyroid cancers diagnosed in North America and Europe and can be divided into invasive and noninvasive subtypes (43). In a recent international study, 109 patients with noninvasive encapsulated follicular variant papillary thyroid carcinoma treated by thyroid lobectomy alone were alive with no evidence of disease at a median of 13 years after treatment. The authors proposed the name “noninvasive follicular thyroid neoplasm with papillary-like nuclear features” for these especially indolent neoplasms and suggested that they not be considered cancers. This new classification would affect more than 45,000 patients annually worldwide (44).

Compared with other tumors, papillary thyroid carcinoma has a low mutation density, which is consistent with its relatively indolent behavior. Kondo and colleagues (45) demonstrated mutually exclusive clonal mutations in papillary thyroid carcinoma (BRAF, 60%; RAS, 15%; and chromosomal rearrangements involving RET and NTRK1, 5%–40%), follicular thyroid carcinoma (RAS, 40%–55%; and rearranged PPARG1, 25%–60%), poorly differentiated thyroid carcinoma (BRAF, 0%–13%; RAS, 20%–30%; CTNNB1, 0%–5%; and TP53, 17%–40%), and anaplastic thyroid carcinoma (BRAF, 10%–35%; RAS, 20%–60%; CTNNB1, 66%; and TP53, 67%–90%).

A recent TCGA study of 496 papillary thyroid carcinomas identified new mutations in EIFIAX, PPM1D, CHEK2 genes and novel chromosomal rearrangements of BRAF, RET, NTRK, and ALK, thereby reducing the number of unknown driver mutations from 25% to 3.5% (3, 46). It is hoped that molecular targeted therapeutics can be developed to effectively target these newly discovered driver mutations. TCGA study investigators were able to divide papillary thyroid carcinomas into three molecular subtypes with mutually exclusive mutations and variable degrees of differentiation: (i) recurrent mutations in genes, with the most prominent being BRAF (59.7%), NRAS, and HRAS, (14.0%); (ii) gene fusions of BRAF, RET, PPARG, NTRK1, NTRK3, ALK, LTK, MET, FGFR2, and THDA (15.3%); and (iii) somatic copy number alterations. The investigators also found that BRAF^V600E-mutated papillary thyroid carcinomas signal preferentially through the MAPK pathway, whereas RAS-mutated papillary thyroid carcinomas signal through either the MAPK pathway or the PI3K pathway (3). These findings support the reclassification of papillary thyroid carcinomas to more accurately define the relationship between genotype and phenotype.

Recent cytologic techniques, based either on proprietary gene expression classification or next-generation sequencing of a panel of oncogenes, have sharpened the diagnosis of inconclusive fine-needle aspirates of thyroid nodules (47, 48). Also, phase III prospective, randomized, placebo-controlled clinical trials of the molecular targeted therapeutics sorafenib and lenvatinib have shown significant improvement in PFS, but not OS in patients with advanced differentiated thyroid carcinoma (Table 1; refs. 49, 50). The FDA has approved both drugs for the treatment of advanced differentiated thyroid carcinoma.

Thyroid lobectomy, or total thyroidectomy, with or without resection of cervical lymph node compartments, is the treatment for primary thyroid cancer, the specific operation depending on tumor histology and the extent of disease. Postoperative adjuvant ¹³¹I-MIBG is administered to patients with a high risk of recurrence. Measurement of thyroglobulin in the postoperative period is useful in detecting persistent or recurrent disease.

Less than 5% of thyroid cancers are medullary thyroid carcinomas, the majority of which are sporadic; however, 25% are hereditary and present as the major part of the MEN syndromes, MEN2A and MEN2B. The tumor cells secrete the polypeptide calcitonin, which serves as an excellent tumor marker for detecting recurrent or persistent disease following thyroidectomy, and especially for determining disease progression, as judged by the rate at which the level of serum calcitonin doubles over time. Approximately 50% of sporadic medullary thyroid carcinomas have somatic RET mutations, and a lesser number have RAS mutations. Virtually all hereditary medullary thyroid carcinomas have germline RET mutations, and there is a strong genotype–phenotype relationship in patients with MEN2A and MEN2B, not only regarding the range of disease expression, but also the severity of disease. On the basis of phase III prospective, randomized, placebo-controlled trials that showed significant improvement in PFS, the FDA approved vandetanib and cabozantinib for the treatment of patients with advanced medullary thyroid carcinoma (Table 1; refs. 51, 52).

The treatment of patients with metastatic thyroid cancer has evolved since 1974 when the FDA approved doxorubicin for the treatment of advanced thyroid cancer (53). Unfortunately, most chemotherapeutics, whether administered alone or in combination with other agents, have shown low response rates of short duration; understandably, the FDA has approved no other chemotherapeutic agent since doxorubicin. Until recently, suppression of thyroid-stimulating hormone and the administration of radioactive iodine were the primary treatments for patients with metastatic differentiated thyroid carcinoma. For patients who failed this therapy, and for patients with advanced medullary thyroid carcinoma, poorly differentiated thyroid carcinoma, or anaplastic thyroid carcinoma, chemotherapy, with or without radiotherapy, was the only option. The advent of molecular targeted therapeutics has opened a new era of therapy for patients with advanced differentiated and medullary thyroid carcinoma. Even though the new drugs provide no OS benefit, they have significantly improved PFS, and often a patient's cancer will remain in remission for several months and sometimes years.

In this CCR Focus, Raue and Frank-Raue provide a succinct overview of the histologic types of thyroid cancer (54). In differentiated thyroid carcinomas, they define patients with low-, intermediate-, and high-risk disease, stressing the importance of dynamic risk assessment, based on several variables, to define both an initial treatment plan for primary thyroid cancer and a strategy to evaluate and treat patients with recurrent or persistent disease following thyroidectomy (55).

Pancreatic neuroendocrine tumors arise from the islets of Langerhans and range from islet cell hyperplasia or adenomas to poorly differentiated neuroendocrine carcinomas. Less than 5% of all pancreatic tumors are neuroendocrine in origin, and their incidence, 0.4 cases per 100,000, has increased over the last three decades, partly due to the aging population and to their incidental detection on imaging studies (56). The prevalence of pancreatic neuroendocrine tumors at autopsy is approximately 10% (57, 58). Sporadic pancreatic neuroendocrine tumors appear around the sixth or seventh decade of life, but hereditary tumors develop at a younger age. Most pancreatic neuroendocrine tumors are nonfunctional and malignant, and over half of them have metastasized at the time of diagnosis. Nonfunctional pancreatic neuroendocrine tumors have a worse prognosis compared with functional tumors, and among the functional tumors, the prognosis varies according to the histologic subtype (59). The World Health Organization grading system for neuroendocrine carcinomas (G1, G2, or G3) is based on an increasing Ki67 index or mitotic count (<2% to >20%), TNM stage, and functional activity (60).

Larsson and colleagues' report that the MEN1 gene is located on chromosome 11, and is most likely a tumor suppressor gene, was confirmed by Chandrasekharappa and colleagues, who cloned the MEN1 gene and named the gene product menin (61, 62). No MEN1 mutation is found in 10% to 30% of typical MEN1 kindreds (probably the result of current DNA sequencing strategies), and 10% of the mutations are de novo, there being no family history (63). MEN1 is also the most common mutation in sporadic pancreatic neuroendocrine tumors, followed by mutations in DAXX/ATRX (death domain–associated protein/mental retardation syndrome X-linked genes) and the mTOR pathway (64). There is no genotype–phenotype relationship in patients with MEN1, unlike MEN2, where specific RET mutations are associated with clinical phenotypes. Furthermore, unlike MEN2, genetic screening in kindreds with MEN1 does not lead to a recommendation regarding medical or surgical intervention.

The primary treatment of pancreatic neuroendocrine tumors is surgical resection, although the procedure fails to cure most patients with malignant tumors. Cytoreductive surgery for bulky regional disease or accessible metastases may reduce morbid symptoms caused by hormone excess (65). Arterial embolization, chemoembolization, radioembolization, and radiofrequency ablation may be effective in some patients with hepatic metastases (66–68).

In this CCR Focus, Maxwell and colleagues provide an excellent overview of the diagnosis and treatment of pancreatic neuroendocrine tumors, stressing the importance of the clinical trials that led to FDA approval of effective therapeutics for the treatment of advanced pancreatic neuroendocrine tumors (69). Until recently, streptozotocin was the only drug approved by the FDA for the treatment of patients with advanced disease; however, the effectiveness of this agent alone, or combined with other chemotherapeutics, remains controversial (70, 71). The majority of pancreatic neuroendocrine tumors have G protein–coupled somatostatin receptors (SSTR), and a reduction in receptor density is associated with increased tumor dedifferentiation (72). Somatostatin acts through a family of five receptors (SSTR1–SSTR5).

Somatostatin analogues were introduced to control symptoms associated with the excess secretion of hormones from pancreatic neuroendocrine tumors. The short- and long-acting synthetic somatostatin analogues octreotide and lanreotide have a strong binding affinity for SSTR2, SSTR3, and SSTR5 (expressed in the brain, anterior pituitary gland, and gastrointestinal tract; SSTR2 is also expressed in the pancreas; ref. 73). Both drugs have shown efficacy in clinical trials of patients with midgut neuroendocrine malignancies (Table 1; refs. 74, 75). In the PROMID study, a long-acting release octreotide, compared with placebo, showed a significantly prolonged time to recurrence and a marked risk reduction in tumor progression (74). In the CLARINET trial, lanreotide, compared with placebo, showed an improved PFS (75). On the basis of the PROMID and CLARINET trial results, the FDA approved octreotide and lanreotide for the treatment of metastatic midgut neuroendocrine tumors. The recently developed somatostatin analogue, pasireotide, binds primarily to all SSTRs except for SSRT4 and is more effective than octreotide in inhibiting GH, IGF1, and ACTH (76). The drug, however, inhibits insulin secretion, and treated patients are at risk for glucose intolerance, which is reversible following cessation of therapy. Recently, the FDA also approved the angiogenesis inhibitor, sunitinib, based on a phase III prospective, randomized trial of patients with advanced pancreatic neuroendocrine tumors that showed improved median PFS in the treatment group compared with placebo (77). The RADIANT I, II, and III trials evaluated the mTOR inhibitor everolimus in patients with pancreatic neuroendocrine tumors (78–80). On the basis of the prolonged PFS, but not OS, compared with placebo (RADIANT III), the FDA also approved everolimus for the treatment of advanced pancreatic neuroendocrine tumors. Even though these new molecular targeted therapeutics have shown efficacy, responses are almost always transient, and the tumor ultimately progresses.

The incidence of symptomatic pituitary tumors in the general population is 1 in 1,000, although recent radiologic and autopsy studies suggest prevalence as high as 22% (81). Functioning endocrine tumors, including prolactinomas (25%–40%), somatotroph adenomas (1%–15%), adrenocorticotrophic adenomas (∼10%), and TSH-producing adenomas (∼1%) are associated with recognizable clinical syndromes. From 5% to 10% of pituitary tumors secrete excess gonadotrophins (LH/FSH), often causing hypogonadism (82). From 30% to 40% of pituitary tumors have no endocrine function, but become symptomatic when they enlarge and compress adjacent structures or destroy normal pituitary cells, causing hypopituitarism (83, 84).

Pituitary tumors are monoclonal, and although they are aneuploid, only 0.2% are malignant (85, 86). Less than 5% of pituitary tumors occur as a benign component of hereditary syndromes, such as MEN1, MEN4, the Carney complex, and familial isolated pituitary adenoma (87–90). The term “MEN4” applies to rare patients with the MEN1 phenotype who have germline mutations in CDKN1B, not MEN1. The Carney complex, characterized by multiple skin lesions, cardiac myxoma, acromegaly, schwannoma, thyroid tumors, and pigmented nodular adrenocortical disease, is caused by germline mutations in CNC1, which encodes the regulatory subunit of the protein kinase A (PRKAR1A). Approximately 65% of patients have mutations in PRKARIA, and 20% have mutations in the putative CNC2 gene located on chromosome 2; however, the specific gene has not been identified. Dominant germline mutations in the aryl hydrocarbon receptor–interacting protein gene, AIP, occur in about 20% of patients with familial isolated pituitary tumors; however, no tumors are found in 80% of family members who carry the mutation, indicating that there are other undiscovered genetic causes of the syndrome.

The majority of pituitary tumors are sporadic and not associated with any known syndrome; however, germline AIP mutations are found in 4% of them, the incidence being higher in children and in young adults with macroadenomas or gigantism. Sporadic tumors with AIP mutations, compared with those without AIP mutations, are more common in males, are larger and invasive, and secrete growth hormone or prolactin. Germline mutations in MEN1 are found in 0.6% to 2.6% of very young patients with isolated pituitary tumors (82, 91). There are few reports of CDKN1B mutations and no reports of PRKAR1A mutations in sporadic pituitary tumors (92).

Although the subject of intense investigation, the role of epigenetic regulation on the pathogenesis of pituitary adenomas is unclear and merits further investigation.

In their CCR Focus article on the genetic background of pituitary adenomas, Caimari and Korbonits discuss nonsyndromic and syndromic pituitary adenomas (93). They also describe a number of somatic mutations associated with sporadic pituitary tumors, but note that the role of these mutations in tumor pathogenesis and progression is unclear.

Although recent discoveries in molecular medicine have led to a deeper understanding of endocrine neoplasia, much remains unknown. Novel molecular targeted therapeutics have improved PFS in patients with some endocrine cancers, yet they have failed to improve OS. With the development of molecular therapeutics that specifically target causative mutations, and the design of combinatorial drug regimens based on an understanding of mechanisms of drug resistance, there is hope for curative therapy. At present, curative therapy depends on the age-old, yet important, principle of establishing the diagnosis of malignancy at an early stage when treatment is curative.

No potential conflicts of interest were disclosed.