Genomic Medicine for Cancer Diagnosis
BENJAMIN L. GORDON, BA,1 BRENDAN M. FINNERTY, MD,1 ANNA ARONOVA, MD,1 AND THOMAS J. FAHEY III, MD2*
1 Research Fellow, Department of Surgery, Weill Cornell Medical College/New York Presbyterian Hospital, NY, New York 2Chief of Endocrine Surgery and Professor of Surgery, Department of Surgery, Weill Cornell Medical College/New York Presbyterian Hospital, NY, New York
Genomic diagnostics in cancer has evolved since the completion of the Human Genome Project and the advancements made in diagnosis and therapy in chronic myelogenous leukemia. Among the diseases to achieve limited success or potentially benefit from diagnostic genetic testing are thyroid cancer, Burkitt’s lymphoma, gastrointestinal stromal tumors, adrenocortical carcinoma, and colorectal cancer. With increased understanding of genomics, genetic tests should improve diagnosis and help guide medical and surgical management. J. Surg. Oncol. @ 2014 Wiley Periodicals, Inc.
KEY WORDS: genetic testing; diagnostic genomics; thyroid cancer; Burkitt’s lymphoma; gastrointestinal stromal tumors
INTRODUCTION
The Human Genome Project was completed in 2003, thirteen years after its launch, having successfully sequenced the human genome [1]. A major finding revealed from the project was that all humans are remarkably similar; in fact, there is only approximately 0.1% genetic difference between individuals [2]. Although these genetic differences are evident in the many forms the human body takes, they may be most impactful in understanding the etiology of disease. Most diseases have a genetic component, whether a mutation is inherited or somatic. In addition to developing clearer prognoses and targeted therapies, detection of these genetic alterations can be used as a diagnostic tool for physicians and surgeons. Among the many diseases with a strong genetic basis, cancer remains a prime example of a field that has the potential to benefit greatly from improved diagnostics through genomics.
In many cases, evaluation of cell morphology on microscopic examination is required for diagnosis. For example, a thyroid nodule detected by ultrasound and classified as “indeterminate” after fine needle biopsy can only be accurately diagnosed after surgical resection. Additionally, different thyroid malignancy subtypes require examination not only of nuclear features, but also architecture of the aggregate cells as well (e.g. classic papillary thyroid carcinoma vs. follicular variant). In these cases, sensitive molecular markers would greatly benefit the patient; with a less ambiguous diagnosis pre-operatively, physicians could better guide medical and surgical management.
At the same time, genomic sequencing has advanced greatly in the past decade. Many genetic tests are reasonably quick, taking weeks, even days to complete. While genetic sequencing is still relatively expensive, the costs of testing have reduced rapidly and will likely become even more affordable in the near future. Next-generation methods of sequencing are promising and may decrease cost without compromising speed, accuracy or reproducibility [3]. Overall, the increased speed and availability, and reduced expense of genetic sequencing have made the incorporation of genomics into modern medicine a reality.
With advances in the field, genomics now offers a possible alternative to effectively diagnose certain cancers that possess limitations in morphologic diagnosis. Among the cancers to first benefit from genetic diagnosis was chronic myelogenous leukemia (CML). The Philadelphia
chromosome (the result of a reciprocal translocation between chromosomes 9 and 22) was discovered in the malignant leukocytes of patients with CML in 1961 [4]. Other studies [5,6] later proved that the fusion gene formed by the translocation (BCR-ABL) and the resulting fusion protein was the cause, and not the effect, of CML [7]. Rational drug design led to the development of targeted drug therapy that selectively inhibited the mutant tyrosine kinase produced by the BCR- ABL fusion protein in CML [8]. Currently, we know that greater than 95% of CML patients have this BCR-ABL fusion gene. Patients are often diagnosed with CML based on the detection of this fusion gene via metaphase cytogenetics [9]; the Philadelphia chromosome can also be detected by fluorescence in situ hybridization (FISH) or polymerase chain reaction (PCR) [10]. In fact, response to therapy can be measured by real-time quantitative reverse transcription polymerase chain reaction (RQ-PCR) for the presence of the BCR-ABL fusion gene and possible resistant mutants [9]. Following the success of genomic medicine in accurately diagnosing and treating CML, nearly every cancer field has attempted to discover the hallmark of its disease. Some, like c-myc in Burkitt’s lymphoma [11], have proven more successful than others.
The Philadelphia chromosome in CML serves as the model for the potential of genomics to better diagnose and effectively treat patients. Physicians in general, and surgeons in particular, should be aware of genomic diagnostics because genetic tests have the potential to accurately predict which patients require surgical intervention instead of clinical observation. As detailed in the coming pages, genomics has already begun to impact diagnostic medicine, and, given the large emphasis of genetics in many cancer fields, will continue to influence patient management for the foreseeable future. In the following review, we will discuss a range of
*Correspondence to: Thomas J. Fahey III, MD, 1300 York Avenue, Room A- 1027, New York, 10065 NY. Fax: (212) 746-9948; E-mail: TJFahey@med. cornell.edu
Received 2 June 2014; Accepted 6 August 2014
DOI 10.1002/jso.23778
Published online in Wiley Online Library (wileyonlinelibrary.com).
diseases that highlight the preliminary success, potential benefits and the overall importance of genomics in cancer diagnosis.
GENOMICS IN THYROID CANCER DIAGNOSIS
The limitation of current diagnostics for a surgeon is perhaps best illustrated in the management of thyroid cancer. Ultrasound-guided FNA is the standard of care for diagnosis; however, its overall accuracy is far from perfect. Even when indeterminate nodules are classified as malignant, it still only has a sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of 95%, 48%, 68%, and 89%, respectively [12,13]. When indeterminate nodules are classified as “indeterminate”, the sensitivity, specificity, PPV, and NPV of FNA are 78%, 60%, 59%, and 79%, respectively [14]. These performance indices are based on the limitations of the Bethesda classification to accurately assess a cytologic specimen’s risk of malignancy [15]: Bethesda I- non- diagnostic (1-4% risk); Bethesda II- benign (<1% risk); Bethesda III- follicular lesion or atypia of undetermined significance (5-25% risk); Bethesda IV- follicular neoplasm or suspicious for follicular neoplasm (20-30% risk); Bethesda V- suspicious for malignancy (50-75% risk); and Bethesda VI- malignant (100% risk).
While the management of Bethesda II and VI lesions is observation and total thyroidectomy, respectively, it is less clear how to treat patients with Bethesda III, IV, and V lesions [16]. Currently, to obtain formal diagnosis, most Bethesda III and IV lesions warrant diagnostic lobectomy. While Bethesda V lesions usually warrant total thyroidectomy, diagnostic lobectomy is commonly pursued as a more conservative measure. Regardless, up to 30% of FNA biopsies will yield indeterminate results (Bethesda III, IV, and V), but, only 10-40% of these nodules prove to be malignant- this gap in diagnostic accuracy thus has a significant impact on surgical management [17,18].
Fortunately, research in thyroid cancer has revealed genomic alterations that differentiate benign and malignant tumors. The most prevalent is the BRAF mutation, as well as the RET/PTC translocation, whose constitutive activation upregulates namely the MAP-kinase and PI3K-AKT cascades. These pathways have a central role in cell growth, proliferation, and apoptosis, as well as expression of proteins essential for thyroid function, such as the sodium-iodide symporter and thyroid peroxidase [19]. Additional genetic alterations include RET point
mutations (associated with medullary carcinoma and MEN2A/2B), PAX8-PPARg rearrangements (associated with follicular carcinomas), PI3K-AKT pathway gene amplifications (associated with anaplastic carcinoma), and tumor suppressor PTEN deletions (associated with follicular carcinoma in Cowden syndrome) [20,21]. Identification of these genomic alterations has ultimately yielded several molecular panels that improve the diagnostic accuracy of FNA, specifically, evaluating BRAF, RAS, RET/PTC, galectin-3, PAX8/PPARg, gene expression classifiers, and microRNA expression profiles, that are both sensitive and specific for diagnosing a malignant lesion on an indeterminate FNA sample [22-25].
One thyroid cancer mutation panel tests for the most common genetic alterations, including BRAF V600E, NRAS codon 61, HRAS codon 61, and KRAS codons 12/13 point mutations and RET/PTC1, RET/PTC3, and PAX8/PPARg rearrangements [24,26]. A large prospective study found this mutational panel to improve the PPVof detecting thyroid cancer in indeterminate nodules from 50-75% to 87-95%, thus eliminating the need for a two-step surgical approach for cancer (i.e. lobectomy followed by completion thyroidectomy) in patients with a positive result; instead, patients can proceed directly to total thyroidectomy. However, the false negative rates were 5%, 14%, and 28% for Bethesda III, IV, and V lesions, respectively; thus a negative result may still warrant a diagnostic lobectomy, particularly for Bethesda IV and V nodules [18].
Given the limitations of mutation testing, some investigators turned to looking for differences in gene expression to try to discriminate benign and malignant thyroid tumors [27-29]. The first commercially available gene- expression classifier, Afirma (Veracyte, South San Francisco, CA), measures the mRNA expression of 167 genes using the Affymetrix microarray platform to differentiate benign from malignant disease with higher sensitivity and negative predictive value than the mutation panel (Table I) [30]. The Afirma test has been clinically validated in a multi-center prospective trial that confirmed an overall sensitivity of 92% and specificity of 52% for correctly identifying indeterminate nodules as suspicious for malignancy [31]. Furthermore, the negative predictive values for Bethesda III, IV, and V nodules were 95%, 94%, and 85%, respectively. Therefore, in the setting of an indeterminate nodule and a negative Afirma test, clinicians can consider close observation instead of diagnostic lobectomy.
The Thymira test (Prolias, New York, NY) evaluates four differentially expressed microRNAs (miR-222, miR-328, miR-197, and miR-21) to
| Study | Cancer | Biomarker | Biospecimen | Sensitivity | Specificity | Ref |
|---|---|---|---|---|---|---|
| Soon et al. (2009) | Adrenocortical carcinoma | IGF-2 profiling and Ki-67 index | Tissue | 96% | 100% | [66] |
| Soon et al. (2009), Patterson et al. (2011) | Adrenocortical carcinoma | miR-483-5p | Tissue | 80% | 100% | [71,72] |
| Dave et al. (2006) | Burkitt's lymphoma | Microarray of 2524 genes | Serum | 100% | — | [31] |
| Imperiale et al. (2014) | Colorectal cancer | KRAS mutation, NDRF3 & BMP3 methylation, b-actin profiling | Feces | 92% | — | [74] |
| Hirota et al. (1998), Novelli et al. (2010) | Gastrointestinal stromal tumors | KIT | Tissue | 95-97% | 90-100% | [45,49] |
| Wang et al. (2013), Novelli et al. (2010), West et al. (2004) | Gastrointestinal stromal tumors | DOG1 | Tissue | 96-99% | 99-100% | [47,49,58] |
| Li et al. (2011) | Lung cancer | miR-21 | Serum | 79% | 100% | [86] |
| Yurkovetsky et al. (2010) | Ovarian cancer | Panel of CA-125, HE4, CEA, VCAM-1 | Serum | 86-95% | 98% | [87] |
| Goonetilleke et al. (2007) | Pancreatic cancer | CA 19-9 | Serum | 70-90% | 68-91% | [88] |
| Liu et al. (2012) | Pancreatic cancer | Panel of miR-20a, miR-21, miR-24, miR-25, miR-99a, miR-185, miR-191 | Serum | 89-94% | 93-100% | [89] |
| Laxman et al. (2008) | Prostate cancer | Panel of PCA3, SPINK1, GOLPH2, TMPRESS2:ERG | Urine | 66% | 76% | [90] |
| Keutgen et al. (2012) | Thyroid Cancer | Thymira: panel of miR-21, miR- 197, miR-222, miR-328 | Tissue | 100% | 95% | [25] |
| Alexander et al. (2012) | Thyroid cancer | Afirma: 167 gene panel | Tissue | 92% | 52% | [28] |
differentiate between malignant and benign indeterminate thyroid lesions [25]. The test applies a non-linear predictive model (Support Vector Model with a radial basis kernel) using the expression of these genes to output a benign or malignant result. The model was derived by correlating the expression levels of these four microRNAs in 29 indeterminate FNA lesions with the accompanying surgical pathology diagnosis. When Hurthle cell lesions were excluded, it was able to predict malignant tumors on a validation set of 72 prospectively collected indeterminate FNA lesions with 100% sensitivity and 95% specificity. Therefore, using this model, total thyroidectomy can be recommended in the setting of a positive result, and observation may be considered in the setting of a negative result.
Lastly, Next-Generation Sequencing (NGS) has been explored as a diagnostic tool for thyroid cancer mutation analysis. One study used NGS to target 12 mutated oncogenes and demonstrated nearly 100% accuracy and sensitivity of detecting mutations with only 10 ng of DNA obtained from FNA specimen [32]. Point mutations were found in 30- 86% of thyroid nodules with different subtypes of cancer, and only 6% of benign nodules. Another study evaluated NGS of FNA specimen with indeterminate results and found 71% sensitivity and 92% NPV for diagnosing thyroid cancer [33]. While these results need to be confirmed in a larger series, NGS appears to represent a viable option to improve diagnostic accuracy pre-operatively.
Genomic panels have become integrated into the surgeon’s algorithm for diagnosing thyroid cancer (Figure 1). At our institution, we consider molecular analysis for Bethesda III, IV, and some V lesions in order to guide clinical observation versus operative management. However, before these panels become a standard in thyroid cancer diagnostics, the overall clinical utility, including cost-analyses for these tests, must be elucidated.
GENETIC PROFILING IN BURKITT’S LYMPHOMA AND DIFFUSE LARGE-B-CELL LYMPHOMA
The development of microarray and genetic sequencing in differentiating Burkitt’s lymphoma from diffuse large-B-cell lymphoma (DLBCL) demonstrates the power of genomics in accurately diagnosing
Bethesda I (non- diagnostic)
Repeat FNA (3 months)
Non-diagnostic
Bethesda II (benign)
Observe: Serial US Tg, TgAb
Bethesda III (follicular atypia)
Diagnostic Lobectomy
Benign
FNA
Molecular Analysis
Non-Diagnostic
Bethesda IV (follicular neoplasm)
Malignant
Bethesda V (suspicious for malignancy)
> 1cm
Total thyroidectomy
Bilateral CND
Bethesda VI (malignancy)
<1cm
Ipsilateral CND
disease and thus, enabling appropriate treatment. Because Burkitt’s lymphoma and DLBCL share many morphologic and immunophenotypic features, morphology and immunostaining are often inadequate to conclusively diagnose one lymphoma over the other [34]. Furthermore, the t(8;14) translocation classically seen in Burkitt’s Lymphoma [35-37] is also present in 5-10% of DLBCL [38]. This alone is not problematic, but coupled with the fact that DLBCL is approximately 20 times more common than Burkitt’s lymphoma [39], the issue of accurate diagnosis of the appropriate lymphoma arises.
Diagnosis between DLBCL and Burkitt’s lymphoma is extremely important because each requires different treatment. DLBCL is typically treated with low-dose chemotherapy regimens like CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone) [40], which is much less effective against Burkitt’s lymphoma [41]. Burkitt’s lymphoma often requires more intense chemotherapy regimens [42,43]. Thus, diagnosis is critical in establishing the appropriate therapy.
Dave et al. examined the utility of gene-expression profiling in differentiating Burkitt’s lymphoma from DLBCL. Using a microarray of 2524 unique genes expressed differentially in different subtypes of non- Hodgkins lymphoma, the investigators were able to accurately identify all cases of Burkitt’s lymphoma agreed upon by a panel of expert pathologists [34]. Additionally, 17% of the cases that the gene- expression profiling labeled as Burkitt’s lymphoma were not detected by pathologists; these specimens shared very similar genetic features to other Burkitt’s lymphoma specimens, suggesting that these cases were Burkitt’s lymphoma unable to be diagnosed by traditional morphologic microscopic examination [34]. Overall, genetic profiling through microarray was able to more accurately diagnose patients with Burkitt’s lymphoma and DLBCL than the current pathological methods. This reproducible genetic test has the potential to save lives and promote quality of life by ensuring that patients are diagnosed, and subsequently, treated appropriately.
DEFINING TUMOR TYPES: C-KIT AND GASTROINTESTINAL STROMAL TUMORS
The discovery of dysfunctional KIT in gastrointestinal stromal tumors (GISTs) further exhibits how molecular markers can elucidate tumor types often indistinguishable by other pathological methods. Prior to this breakthrough, the early 1990s marked a perplexing time in the diagnosis of spindle cell and epithelioid tumors that originated from the stromal/ mesenchymal components of the gastrointestinal tract, particularly GISTs. Newman et al. attempted to define these tumors by lines of differentiation, grouping them into neural, smooth muscular, bidirectional differentiation, and null phenotype categories [44]. However, there was much variability in the relative proportions of these different groups in subsequent studies, leaving ambiguity in this classification system [45]. While CD34 immunostaining emerged as a promising diagnostic tool capable of differentiating spindle cell and epithelioid lesions from schwannomas and leiomyomas [46], only 60-70% of GISTs are CD34-positive in addition to a number of schwannomas and leiomyomas that are CD34-positive as well [45]. By the mid-1990 s, diagnosis of GISTs stood at an impasse, with physicians either simply grouping together GISTs, schwannomas and leiomyomas under one heading, or attempting to exclude true leiomyomas and schwannomas from other gastrointestinal mesenchymal tumors, albeit doing so poorly without a sensitive marker [45].
In 1998, Hirota et al. examined gain-of-function mutations in c-kit, a proto-oncogene that encodes a type III receptor tyrosine kinase (KIT) [48]. Approximately 50% of the cDNA clones from each GIST exhibited mutations in the regions between the transmembrane and tyrosine kinase domains of c-kit, leading to constitutive activation of the c-kit receptor tyrosine kinase without the KIT ligand, stem cell factor [48]. This continuous activation led to the overexpression of downstream signaling pathways responsible for tumorigenesis, chiefly proliferation, cell growth,
Journal of Surgical Oncology
cell differentiation and apoptosis [49-51]. Moreover, KIT was highly sensitive and specific for GISTs on immunohistochemistry, expressed in 94% of GISTs and in no leiomyomas and schwannomas [48]. Finally, GISTs had a reliable molecular marker to differentiate them from schwannomas and leiomyomas. Years later, KIT remains a highly sensitive (95-97%) and specific (90-100%) biomarker [48,52].
Mutations in alpha-type platelet-derived growth factor receptor (PDGFRA), a highly homologous receptor tyrosine kinase protein to KIT [53], have also been identified as a cause of GIST tumorigenesis. While mutations in PDGFRA are far less common than in c-kit [48], PDGFRA is mutated in 35-75% of wild-type KIT tumors [54,55]. Pre- operative FNA enables mutational analysis for c-kit and PDGFRA, enabling diagnosis before surgery [56]. Mutational analysis in GISTs is not only important for diagnosis, but can also guide treatment. The efficacy of imatinib and sunitinib, both of which target KIT and PDGFRA, directly relates to mutations detected in patient’s tumors and even depends on different mutations within the same gene [57].
More recently, it has been demonstrated that no c-kit or PDGFRA mutation exists in 5-15% of GISTs [58-60] and that 5-10% of GISTs do not immunostain for CD117 (KIT) [53]. DOG1, a calcium-dependent receptor-activated chloride ion channel protein [50], has been shown to be a potential effective complementary tool in diagnosing GISTs. On immunochemistry, 96-99% of GISTs stain for DOG1, with a specificity of 99-100%, even staining most KIT-mutation-negative GISTs [50,52,61]. PKC-0, a member of protein kinase C [62], while less sensitive (91-92%) and specific (68-80%) for diagnosis of GIST, immunostains in 75-88% of KIT-negative GISTs [50,63]. Both DOG1 and PKC-0 pose as promising biomarkers, especially in the diagnosis of KIT-negative GISTs.
GENOMIC MARKERS IN ADRENOCORTICAL CARCINOMA
In many cancer types, surgical resection and pathologic evaluation are the sole methods for determining malignancy. Many common tumors are subject to such invasive diagnosis, including thyroid cancer as noted previously. Rare tumors can also pose a similar diagnostic dilemma that can currently only be settled by surgical resection. With these rare tumors, small sample sizes have precluded rapid progress in molecular diagnostics. One such cancer type is adrenocortical carcinoma (ACC)- a rare, but deadly malignancy affecting 1-2 patients per million per year, this cancer carries an overall 5-year survival rate of 19- 45% [64,65]. This is largely due to its aggressive nature coupled with deficient tools for early diagnosis.
Several groups are looking at genomic markers for early diagnosis of ACC in order to distinguish it from its more common, benign counterpart adrenal adenoma (AA) using various genomic approaches. Gene expression profiling of adrenocortical tumors has identified several diagnostic molecular markers. The most common of these is insulin-like growth factor-2 (IGF-2), which exerts its actions through the IGF-1 receptor (IGF-1R). Overexpression of IGF-2 at both the mRNA and protein levels has been found to be significantly higher in adrenocortical carcinomas compared to benign adenomas. One common molecular mechanism to account for this is loss of heterozygosity at the IGF-2 locus of 11p15 [66]. Regardless of its underlying molecular perturbation, however, IGF-2 overexpression has been shown to be over 100-fold higher in 60-90% of ACC% compared to benign or normal adrenal tissue [67,68]. Furthermore, when coupled with Ki-67 index, Soon et al. found 96% sensitivity and 100% specificity in discriminating malignant from benign adrenocortical tumors [69]. Multiple other studies have similarly found molecular profiles that seem to discriminate well between ACC and AA on pathologic specimens. Such differentially expressed genes include vascular endothelial growth factor (VEGF), steroidogenic factor-1 (SF-1), and glucocorticoid receptor (GR) [69-71], and have each been found to accurately distinguish ACC from AA. However, at this
point, these molecular markers are limited to assisting in the accurate characterization of surgical specimens, as most adrenal lesions are not pre-operatively biopsied.
Comparative genomic hybridization (CGH) has been used to show that ACC is generally associated with significant chromosomal losses and gains compared to benign tissues [66]. A recent study comparing 138 samples (52 ACCs and 86 AAs) found chromosomal alterations in 44% of ACCs versus 10% of AAs [72]. Specifically, gains were found at chromosomes 5, 7, 12, 16,19, and 20 and losses at 13 and 20. Combining DNA copy number estimates at 6 loci (5q, 7p, 11p, 13q, 16q, and 22q), the authors developed a diagnostic signature that demonstrated 100% sensitivity and 83% specificity in discriminating ACC from AA on an independent validation cohort of 79 tumors [72].
Genome-wide expression studies and comparative genomic hybridization (CGH) have also been used to define dysregulated molecular pathways in ACC. A meta-analysis by Szabo et al. showed that cell cycle, retinoic acid signaling, cholesterol and lipid metabolism, toll-like receptor 4, the complement system, and antigen presentation were all dysregulated in ACC [73].
Other genomic methods employed for molecular diagnosis of ACC have included microRNA profiling. MiR-195 down-regulation and miR- 483-5p overexpression have been observed in several studies, with miR- 483-5p demonstrating a high diagnostic accuracy for distinguishing benign from malignant adrenal tumors with 80% sensitivity and 100% specificity [74,75]. While these genomic profiles are constantly improving our accuracy in pathologic diagnosis and aiding our understanding of the molecular alterations involved in tumorigenesis, their pre-operative diagnostic capabilities have yet to be exploited.
Recently, however, microRNAs have recently been suggested for use as serum biomarkers. Specifically, miR-34a and miR-483-5p were found at significantly higher levels in serum of patients with ACC compared to those with AA [76]. Although promising, this approach needs further study and validation before potential implementation in clinical practice.
In general, there has been considerable progress in understanding the molecular pathogenesis of ACC. Yet, there is still significant heterogeneity of findings prohibiting routine implementation for diagnostic purposes. This is in part due to the heterogeneity of the tumors themselves, and partly due to the rarity of this disease, making large-scale studies difficult. Nonetheless, as the bioinformatics field advances, we will be better equipped to integrate these various genomic methods to determine genomic markers for diagnosis of such tumors as ACC.
GENOMICS IN STOOL ANALYSIS
The ultimate utility of genomics in cancer diagnosis lies in its potential ability for early detection. Recent literature suggests there is a role for genomics analysis of fecal samples for detection of colorectal and pancreatic tumors [77-81]. In one of the largest trials of its kind to date, nearly 10,000 patients of average colorectal cancer risk underwent DNA testing (including KRAS mutation, aberrant NDRF3 & BMP3 methylation, and b-actin analyses), fecal immunohistochemistry (FIT), and screening colonoscopy [77]. The study found that the DNA testing was more sensitive than FIT for detecting colorectal tumors (92% vs. 74%, P=0.002) and precancerous lesions (42% vs. 24%, P<0.001). These results suggest that mutational analyses may be a successful non- invasive adjunct to colorectal cancer screening diagnostics.
In addition to mutational analyses, initial reports have shown that fecal microRNA extraction and expression is reproducible and effective at detecting differentially expressed microRNAs in patients with and without colorectal and pancreatic tumors [78-81]. These studies have identified specific microRNAs that form a neoplastic signature, namely miR-21 and miR-106a in colorectal cancer, and miR-181b and miR-210 in pancreatic cancer. These microRNA expression profiles will have to be confirmed and expanded upon in a larger cohort in order to evaluate their accuracy and sensitivity as a screening marker for early cancer detection.
COST-ANALYSIS OF DIAGNOSTIC GENETIC TESTING
Very little work has been done to investigate the cost effectiveness of genetic testing in diagnosis, with most investigations examining the cost benefit of genomics in prognosis, primarily recurrence, and metastasis [82-84]. One study examining diagnosis of familial paraganglioma syndrome (FPS) found that patients saved greater than $2200 over a six-year period by genetically screening family members of FPS patients, citing early detection as preventing morbidity [85]. However, the sample in this study was very small. Expanding cost- analysis studies to the genetic testing of other cancers will help elucidate the long-term benefits of genomic diagnosis.
LIMITATIONS TO GENOMIC DIAGNOSTICS
Three issues arise when contemplating the current and potential use of genomics in cancer diagnosis. The first is a question of utility- chiefly, will genetic testing be a useful tool for physicians in evaluating their patients? In a study of 310 indeterminate thyroid nodule FNA biopsies, pre-operative BRAF mutation screening only had a 15% sensitivity of detecting malignancy in indeterminate nodules. Moreover, mutation positivity did not impact initial surgical management: 12 out of 13 patients with BRAF mutation in an indeterminate nodule initially underwent total thyroidectomy due to previously identified worrisome cytologic features, and thus, only one patient’s surgical management would have been altered [16]. Given these data, BRAF mutation screening alone may not be a robust diagnostic tool for pre-operative management. While other genetic tests have proven more useful than BRAF screening, the worry with genomics is that it only provides minimal benefit, and at worst, possible harm through misdiagnosis, to the patient. Until physicians are confident that molecular markers are reliable and accurate diagnostic tools for a given cancer, and feel comfortable basing treatment on the results of a genetic test, the success of genomic medicine will be impeded.
The next two limitations to genomic diagnosis are intertwined. One, even in a specific cancer, there is often broad heterogeneity of tumors. Simply put, the same cancer may arise in two different patients as the result of two different mutations; the underlying pathogenesis of a given cancer type may diverge between two tumors on a molecular level. Next-Generation Sequencing may be able to help elucidate these molecular mechanisms, revealing shared genetics of heterogeneous tumors labeled under one heading or separating these tumors into different categories altogether. This leads to the last limitation- NGS is in its infant stages. While NGS is currently able to generate large amounts of data on a specimen(s) and churn out differential expression profiles, integrated pathway analyses still lag behind NGS. Only with improved pathway analyses can we fully understand the implications of NGS and target driving forces behind oncogenesis.
CONCLUSION
Even given the current limitations of cancer diagnosis through genomics, the future remains promising. Since we do not fully understand all of the of the signal transduction pathways involved in oncogenesis, genomics has the potential to identify currently imperceptible mechanisms contributing to the molecular basis of cancer. Much like morphology has helped define certain variants of thyroid cancer, genomics can define additional distinct variants and subtypes of a given cancer based on specific pathway involvement. Only when these mechanisms are thoroughly understood can genomic analysis be ubiquitously used as a successful adjunct for cancer diagnostics.
REFERENCES
1. Collins FS, Green ED, Guttmacher AE, et al .: A vision for the future of genomics research. Nature 2003;422:835-847.
2. Tishkoff SA, Kidd KK: Implications of biogeography of human populations for ‘race’ and medicine. Nat Genet 2004;36:S21-S27.
3. Ong FS, Jin JC, Das K, et al .: Translational utility of next-generation sequencing. Genomics 2013;102:137-139.
4. Nowell PC, Hungerford DA: Chromosome studies in human leukemia II: Chronic granulocytic leukemia. J Natl Cancer Inst 1961;27:1013-1035.
5. Daley GQ, Van Etten RA, Baltimore D: Induction of chronic myelogenous leukemia in mice by the P210bcr/abl gene of the Philadelphia chromosome. Science 1990;247:824-830.
6. Kelliher MA, Mclaughlin J, Witte ON, et al .: Induction of a chronic myelogenous leukemia-like syndrome in mice with v-abl and BCR/ ABL. Proc Natl Acad Sci USA 1990;87:6649-6653.
7. Koretzky GA, The legacy of the Philadelphia chromosome. J Clin Invest 2007;117:2030-2032.
8. Druker BJ, Tamura S, Buchdunger E, et al .: Effects of a selective inhibitor of the Abl tyrosine kinase on the growth of Bcr-Abl positive cells. Nat Med 1996;2:561-566.
9. Luu MH, Press RD: BCR-ABL PCR testing in chronic myeloge- nous leukemia: Molecular diagnosis for targeted cancer therapy and monitoring. Expert Rev Mol Diagn 2013;13:749-762.
10. Kawasaki ES, Clark SS, Coyne MY, et al .: Diagnosis of chronic myeloid and acute lymphocytic leukemias by detection of leukemia- specific mRNA sequences amplified in vitro. Proc Natl Acad Sci USA 1988;85:5698-5702.
11. Lombardi L, Newcomb EW, Dalla-Favera R: Pathogenesis of Burkitt lymphoma: Expression of an activated c-myc oncogene causes the tumorigenic conversion of EBC-infected human B lymphoblasts. Cell 1987;49:161-170.
12. Danese D, Sciacchitano S, Farsetti A, et al .: Diagnostic accuracy of conventional versus sonography-guided fine-needle aspiration biopsy of thyroid nodules. Thyroid 1998;8:15-21.
13. Wang CC, Friedman L, Kennedy GC, et al .: A large multicenter correlation study of thyroid nodule cytopathology and histopathol- ogy. Thyroid 2011;21:243-251.
14. Kleiman DA, Sporn MJ, Beninato T, et al .: Preoperative BRAF (V600E) mutation screening is unlikely to alter initial surgical treatment of patients with indeterminate thyroid nodules: A prospective case series of 960 patients. Cancer 2013;119:1495- 1502.
15. Baloch ZW, LiVolsi VA, Asa SL, et al .: Diagnostic terminology and morphologic criteria for cytologic diagnosis of thyroid lesions: A synopsis of the National Cancer Institute Thyroid Fine-Needle Aspiration State of the Science Conference. Diagn Cytopathol 2008;36:425-437.
16. Cibas ES, Ali SZ: The Bethesda system for reporting thyroid cytopathology. Thyroid 2009;19:1159-1165.
17. Mazzaferri EL: Management of a solitary thyroid nodule. N Engl J Med 1993;328:553-559.
18. Nikiforov YE, Ohori NP, Hodak SP, et al .: Impact of mutational testing on the diagnosis and management of patients with cytologically indeterminate thyroid nodules: A prospective analysis of 1056 FNA samples. J Clin Endocrinol Metab 2011;96:3390- 3397.
19. Hilger RA, Scheulen ME, Strumberg D: The Ras-Raf-MEK-ERK pathway in the treatment of cancer. Onkologie 2002;25:511-518.
20. Nikiforov YE: Thyroid carcinoma: Molecular pathways and therapeutic targets. Mod Pathol 2008;21:S37-S43.
21. Xing M: Molecular pathogenesis and mechanisms of thyroid cancer. Nat Rev Cancer 2013;13:184-199.
22. Bartolazzi A, Orlandi F, Saggiorato E, et al .: Galectin-3-expression analysis in the surgical selection of follicular thyroid nodules with indeterminate fine-needle aspiration cytology: A prospective multicentre study. Lancet Oncol 2008;9:543-549.
23. Sapio MR, Posca D, Raggioli A, et al .: Detection of RET/PTC, TRK and BRAF mutations in preoperative diagnosis of thyroid nodules with indeterminate cytological findings. Clin Endocrinol (Oxf) 2007;66:678-683.
Journal of Surgical Oncology
Gordon et al.
24. Nikiforov YE, Steward DL, Robinson-Smith TM, et al .: Molecular testing for mutations in improving the fine-needle aspiration diagnosis of thyroid nodules. J Clin Endocrinol Metab 2009;94: 2092-2098.
25. Keutgen XM, Filicori F, Crowley MJ, et al .: A Panel of four miRNAs accurately differentiates malignant from Benign indeter- minate thyroid lesions on fine needle aspiration. Clin Cancer Res 2012;18:2032-2038.
26. Cantara S, Capezzone M, Marchisotta S, et al .: Impact of proto- oncogene mutation detection in cytological specimens from thyroid nodules improves the diagnostic accuracy of cytology. J Clin Endocrinol Metab 2010;95:1365-1369.
27. Barden CB, Shister KW, Zhu B, et al .: Classification of follicular thyroid tumors by molecular signature: Results of gene profiling. Clin Cancer Res 2003;9:1792-1800.
28. Finley DJ, Arora N, Zhu B, et al .: Molecular profiling distinguishes papillary carcinoma from benign thyroid nodules. J Clin Endocrinol Metab 2004;89:3214-3223.
29. Finley DJ, Zhu B, Barden CB, et al .: Discrimination of benign and malignant thyroid nodules by molecular profiling. Ann Surg 2004; 240:425-437.
30. Chudova D, Wilde JI, Wang ET, et al .: Molecular classification of thyroid nodules using high-dimensionality genomic data. J Clin Endocrinol Metab 2010;95:5296-5304.
31. Alexander EK, Kennedy GC, Baloch ZW, et al .: Preoperative diagnosis of benign thyroid nodules with indeterminate cytology. N Engl J Med 2012;367:705-715.
32. Nikiforova MN, Wald AI, Roy S, et al .: Targeted next-generation sequencing panel (ThyroSeq) for detection of mutations in thyroid cancer. J Clin Endocrinol Metab 2013;98:E1852-E1860.
33. Le Mercier M, D’Haene N, De Neve N, et al .: Next-generation sequencing improves the diagnosis of thyroid FNA specimens with indeterminate cytology. Histopathology 2014;[Epub ahead of print].
34. Dave SS, Fu K, Wright GW, et al .: Molecular diagnosis of Burkitt’s lymphoma. N Engl J Med 2006;354:2431-2442.
35. Neri A, Barriga F, Knowles DM, et al .: Different regions of the immunoglobulin heavy-chain locus are involved in chromosomal translocations in distinct pathogenetic forms of Burkitt lymphoma. Proc Natl Acad Sci USA 1988;85:2748-2752.
36. Gerbitz A, Mautner J, Geltinger C, et al .: Deregulation of the proto- oncogene c-myc through t (8;22) translocation in Burkitt’s lymphoma. Oncogene 1999;18:1745-1753.
37. Hecht JL, Aster JC: Molecular biology of Burkitt’s lymphoma. J Clin Oncol 2000;18:3707-3721.
38. Kramer MH, Hermans J, Wijburg E, et al .: Clinical relevance of BCL2, BCL6, and MYC rearrangements in diffuse large B-cell lymphoma. Blood 1998;92:3152-3162.
39. Morton LM, Wang SS, Devesa SS, et al .: Lymphoma incidence patterns by WHO subtype in the United States, 1992-2001. Blood 2006;107:265-276.
40. Martelli M, Ferreri AJM, Agostinelli C, et al .: Diffuse large B-cell lymphoma. Crit Rev Oncol Hematol 2013;87:146-171.
41. Butler RD, Hainsworth JD: Optimal therapy for small noncleaved cell lymphoma. Cancer Treat Res 1993;66:65-79.
42. Magrath I, Adde M, Shad A, et al .: Adults and children with small non-cleaved-cell lymphoma have a similar excellent outcome when treated with the same chemotherapy regimen. J Clin Oncol 1996;14: 925-934.
43. Diviné M, Casassus P, Koscielny S, et al .: Burkitt lymphoma in adults: A prospective study of 72 patients treated with an adapted pediatric LMB protocol. Ann Oncol 2005;16:1928-1935.
44. Newman PL, Wadden C, Fletcher CDM: Gastrointestinal stromal tumors: Correlation of immunophenotype with clinicopathologic features. J Pathol 1991;164:107-117.
45. Fletcher CD, Berman JJ, Corless C, et al .: Diagnosis of gastrointestinal stromal tumors: A consenus approach. Int J Surg Pathol 2002;10:81-89.
46. Miettinen M, Virolainen M, Sarlomo-Rikala M: Gastrointestinal stromal tumors: Value of CD34 antigen in their identification and separation from true leoimyomas and schwannomas. Am J Surg Pathol 1995;19:207-216.
47. Chabot B, Stephenson DA, Chapman VM, et al .: The proto- oncogene c-kit encoding a transmembrane tyrosine kinase receptor maps to the mouse W locus. Nature 1988;335:88-89.
48. Hirota S, Isozaki K, Moriyama Y, et al .: Gain-of-function mutations of c-kit in human gastrointestinal stromal tumors. Science 1998; 279:577-580.
49. Joensuu H: Risk stratification of patients diagnosed with gastrointestinal stromal tumor. Hum Pathol 2008;39:1411-1419.
50. Wang C, Jin M, Zou Y, et al .: Diagnostic significance of DOG-1 and PKC-0 expression and c-Kit/PDGFRA mutations in gastrointestinal stromal tumors. Scand J Gastroenterol 2013;48:1055-1065.
51. Gramza AW, Corless CL, Heinrich MC: Resistance to tyrosine kinase inhibitors in gastrointestinal stromal tumors. Clin Cancer Res 2009;15:7510-7518.
52. Novelli M, Rossi S, Rodriguez-Justo M, et al .: DOG1 and CD117 are the antibodies of choice in the diagnosis of gastrointestinal stromal tumours. Histopathology 2010;57:259-270.
53. Miettinen M, Lasota J: Gastrointestinal stromal tumors: Review on morphology, molecular pathology, prognosis, and differential diagnosis. Arch Pathol Lab Med 2006;130:1466-1478.
54. Heinrich MC, Corless CL, Duensing A, et al .: PDGFRA activating mutations in gastrointestinal stromal tumors. Science 2003;299: 708-710.
55. Hirota S, Ohashi A, Nishida T, et al .: Gain-of-function mutations of platelet-derived growth factor receptor « gene in gastrointestinal stromal tumors. Gastroenterol 2003;125:660-667.
56. Willmore-Payne C, Layfield LJ, Holden JA: c-KIT mutation analysis for diagnosis of gastrointestinal stromal tumors in fine needle aspiration specimens. Cancer 2005;105:165-170.
57. Heinrich MC, Maki RG, Corless CL, et al .: Primary and secondary kinase genotypes correlate with the biological and clinical activity of sunitinib in imatinib-resistant gastrointestinal stromal tumor. J Clin Oncol 2008;26:5352-5359.
58. DeMatteo RP, Ballman KV, Antonescu CR, et al .: Adjuvant imatinib mesylate after resection of localised, primary gastrointes- tinal stromal tumour: A randomised, double-blind, placebo- controlled trial. Lancet 2009;373:1097-1104.
59. Demetri GD, von Mehren M, Antonescu CR, et al .: NCCN task force report: Update on the management of patients with gastrointestinal stromal tumors. J Natl Compr Canc Netw 2010;8:S1-S41.
60. Debiec-Rychter M, Wasag B, Stul M, et al .: Gastrointestinal stromal tumours (GISTs) negative for KIT (CD117 antigen) immunoreac- tivity. J Pathol 2004;202:430-438.
61. West RG, Corless CL, Chen X, et al .: The novel marker, DOG1, is expressed ubiquitously in gastrointestinal stromal tumors irrespective of KIT or PDGFRA mutation status. Am J Pathol 2004;165:107-113.
62. Baier G, Telford D, Giampa L, et al .: Molecular cloning and characterization of PKC theta, a novel member of the protein kinase C (PKC) gene family expressed predominantly in hematopoietic cells. J Biol Chem 1993;268:4997-5004.
63. Lee HE, Kim MA, Lee HS, et al .: Characteristics of KIT-negative gastrointestinal stromal tumours and diagnostic utility of protein kinase C theta immunostaining. J Clin Pathol 2008;61:722-729.
64. Kuruba R, Gallagher SF: Current management of adrenal tumors. Curr Opin Oncol 2008;20:34-46.
65. Libé R, Bertherat J: Molecular genetics of adrenocortical tumours, from familial to sporadic diseases. Eur J Endocrinol 2005;153:477- 487.
66. Soon PSH, McDonald KL, Robinson BG, et al .: Molecular markers and the pathogenesis of adrenocortical cancer. Oncologist 2008; 13:548-561.
67. Barlaskar FM, Hammer GD: The molecular genetics of adrenocor- tical carcinoma. Rev Endocr Metab Disord 2007;8:343-348.
68. de Fraipont F, El Atifi M, Cherradi N, et al .: Gene expression profiling of human adrenocortical tumors using complementary deoxyribonucleic acid microarrays identifies several candidate genes as markers of malignancy. J Clin Endocrinol Metab 2005;90: 1819-1829.
69. Soon PSH, Gill aJ, Benn DE, et al .: Microarray gene expression and immunohistochemistry analyses of adrenocortical tumors identify IGF2 and Ki-67 as useful in differentiating carcinomas from adenomas. Endocr Relat Cancer 2009;16:573-583.
70. Tacon LJ, Soon PS, Gill AJ, et al .: The glucocorticoid receptor is overexpressed in malignant adrenocortical tumors. J Clin Endo- crinol Metab 2009;94:4591-4599.
71. Sbiera S, Schmull S, Assie G, et al .: High diagnostic and prognostic value of steroidogenic factor-1 expression in adrenal tumors. J Clin Endocrinol Metab 2010;95:E161-E171.
72. Barreau O, de Reynies A, Wilmot-Roussel H, et al .: Clinical and pathophysiological implications of chromosomal alterations in adrenocortical tumors: an integrated genomic approach. J Clin Endocrinol Metab 2012;97:E301-E311.
73. Szabó PM, Tamási V, Molnár V, et al .: Meta-analysis of adrenocortical tumour genomics data: novel pathogenic pathways revealed. Oncogene 2010;29:3163-3167.
74. Soon PSH, Tacon LJ, Gill AJ, et al .: miR-195 and miR-483-5p identified as predictors of poor prognosis in adrenocortical cancer. Clin Cancer Res 2009;15:7684-7692.
75. Patterson EE, Holloway AK, Weng J, et al .: MicroRNA profiling of adrenocortical tumors reveals miR-483 as a marker of malignancy. Cancer 2011;117:1630-1639.
76. Patel D, Boufragech M, Jain M, et al .: MiR-34a and miR-483-5p are candidate serum biomarkers for adrenocortical tumors. Surgery 2013;154:1224-1228.
77. Imperiale TF, Ransohoff DF, Itzkowitz SH, et al .: Multitarget stool DNA testing for colorectal-cancer screening. N Engl J Med 2014;370:1287-1297.
78. Link A, Balaguer F, Shen Y, et al .: Fecal MicroRNAs as novel biomarkers for colon cancer screening. Cancer Epidemiol Bio- markers Prev 2010;19:1766-1774.
79. Ahmed FE, Ahmed NC, Vos PW, et al .: Diagnostic microRNA markers to screen for sporadic human colon cancer in stool: I. Proof of principle. Cancer Genomics Proteomics 2013;10:93-113.
80. Ren Y, Gao J, Liu J, et al .: Differential signature of fecal microRNAs in patients with pancreatic cancer. Mol Med Rep 2012;6:201-209.
81. Link A, Becker V, Goel A, et al .: Feasibility of fecal microRNAs as novel biomarkers for pancreatic cancer. PLOS One 2012;7:e42933. doi: 10.1371/journal.pone.0042933
82. Blons H, Rouleau E, Charrier N, et al .: Performance and cost efficiency of KRAS mutation testing for metastatic colorectal cancer in routine diagnosis: The MOKAECM study, a nationwide experience. PLOS 2013;8:e68945.doi: 10.1371/journal.pone.0068945
83. Frank M, Mittendorf T: Influence of pharmacogenomic profiling prior to pharmaceutical treatment in metastatic colorectal cancer on cost effectiveness: A systematic review. Pharmaeconomics 2013; 31:215-228.
84. Klang SH, Hammerman A, Liebermann N, et al .: Economic implications of 21-gene breast cancer risk assay from the perspective of an Israeli-managed health-care organization. Value Health 2010;13:381-387.
85. Dziegielewski PT, Knox A, Liu R, et al .: Familial paraganglioma syndrome: Applying genetic screening in otolaryngology. J Otolaryngol Head Neck Surg 2010;39:646-653.
86. Li Y, Li W, Ouyang Q, et al .: Detection of lung cancer with blood microRNA-21 expression levels in Chinese population. Oncol Lett 2011;2:991-994.
87. Yurkovetsky Z, Skates S, Lomakin A, et al .: Development of a multimarker assay for early detection of ovarian cancer. J Clin Oncol 2010;28:2159-2166.
88. Goonetilleke KS, Siriwardena AK: Systematic review of carbohy- drate antigen (CA 19-9) as a biochemical marker in the diagnosis of pancreatic cancer. Eur J Surg Oncol 2007;33:266-270.
89. Liu R, Chen X, Du Y, et al .: Serum miRNA expression profile as a biomarker in the diagnosis and prognosis of pancreatic cancer. Clin Chem 2012;58:610-618.
90. Laxman B, Morris DS, Yu J, et al .: A first-generation multiplex biomarker analysis of urine for the early detection of prostate cancer. Cancer Res 2008;68:645-649.