Utility of PTEN Protein Dosage in Predicting for Underlying Germline PTEN Mutations among Patients Presenting with Thyroid Cancer and Cowden-Like Phenotypes

Abstract

Context:

Thyroid cancer is a major component of Cowden syndrome (CS). CS patients with an underlying PTEN mutation (PTEN^mut+) have a 70-fold increased risk of developing epithelial thyroid cancer. In contrast, less than 1% of sporadic epithelial thyroid cancer patients carry a germline PTEN mutation. Cost-efficient markers capable of shortlisting thyroid cancers for CS genetic testing would be clinically useful.

Objective:

Our objective was to analyze the utility of patient blood phosphate and tensin homolog deleted on chromosome 10 (PTEN) protein levels in predicting germline PTEN mutations.

Design, Setting, and Patients:

We conducted a 5-yr, multicenter prospective study of 2792 CS and CS-like patients, all of whom had comprehensive PTEN analysis. Analysis of PTEN and downstream proteins by immunoblotting was performed on total protein lysates from patient-derived lymphoblast lines. We compared blood PTEN protein levels between PTEN^mut+ patients and those with variants of unknown significance or wild-type PTEN (PTEN^wt/vus).

Main Outcome Measures:

We assessed the utility of PTEN protein levels in predicting germline PTEN mutations.

Results:

Of 2792 CS/CS-like patients, 721 patients had thyroid cancer; 582 of them (81%) had blood PTEN protein analyzed. PTEN germline pathogenic mutations were present in 27 of 582 patients (4.6%). Ninety-six percent (26 of 27) of PTEN^mut+ patients had blood PTEN protein levels in the lowest quartile as compared with 25% (139 of 555) of PTEN^wt/vus patients (P < 0.001). Low blood PTEN levels predicted for PTEN^mut+ cases with a 99.76% negative predictive value (95% confidence interval = 98.67–99.96) and a positive test likelihood ratio of 3.84 (95% confidence interval = 3.27–4.52).

Conclusions:

Our study shows that low blood PTEN protein expression could serve as a screening molecular correlate to predict for germline PTEN mutation in CS and CS-like presentations of thyroid cancer.

Thyroid cancer is the most rapidly rising incident cancer in women and the second most rapidly rising incident cancer in men in the United States. In 2011, according to the American Cancer Society, approximately 48,000 new cases of thyroid cancer were diagnosed and caused mortality in more than 1700 patients, a number that is rising yearly despite aggressive therapy with surgery, radioactive iodine, and TSH suppression with l-T₄. Importantly, 2012 SEER (surveillance epidemiology and end results) data note prevalence of thyroid cancer in more than 496,000 individuals, of which approximately 300,000 are females, with almost 90,000 (20%) having recurrent or residual disease (1). Cancer registry data have also demonstrated a very high familiality and, thus, a strong genetic component. Epithelial thyroid carcinomas are a major component of Cowden and related syndromes characterized by germline mutations in the PTEN tumor suppressor gene, and collectively known as PTEN hamartoma tumor syndromes (PHTS) (2, 3). The syndromes encompassed by PHTS include Cowden syndrome (CS), Bannayan-Riley-Ruvalcaba syndrome, and Proteus-like syndrome (4). These disorders share phenotypic similarities but are clinically distinct, highlighting the protean spectrum of PHTS. Clinical criteria for CS are based on guidelines developed by the International Cowden Consortium (5, 6) and subsequently adopted by the National Comprehensive Cancer Network and include pathognomonic, major, and minor criteria (Supplemental Table 1, published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org) (7). CS is an autosomal dominant disorder characterized by the development of multiple hamartomas, and importantly carcinomas, of the thyroid, breast, endometrium, and kidney (3, 8). Based on early epidemiological studies, nonmedullary thyroid cancer was identified as a major feature and benign thyroid disorders such as thyroid adenoma and nodular hyperplasia were included as minor features toward clinical diagnosis (9). In a recent large prospective study, we confirmed and quantified that germline mutations in PTEN increased the risk of epithelial thyroid cancer by more than 70-fold when compared with that of the general population (2). In contrast, the prevalence of germline PTEN mutations in unselected differentiated thyroid cancer is low (<1%) (10, 11). It is important to efficiently identify PTEN-related thyroid cancers among all presentations of epithelial thyroid cancer because of the increased risks of extrathyroidal cancers in PHTS, namely 85% lifetime risk of breast cancer, 28% of uterine cancer, and 33% of renal cancer (3).

In addition to complete loss of function, alterations of tumor suppressor genes in cancer can lead to partial loss-of-function, gain-of-function, or dominant-negative phenotypes. Inheriting a mutation in PTEN predisposes to cancer, but the inactivation of phosphate and tensin homolog deleted on chromosome 10 (PTEN) has been shown to occur via multiple diverse mechanisms across the different tumor types (12–15). The germline mutational spectrum in PHTS is broad with mutations scattered across the PTEN coding region (16, 17). Correlations between specific germline PTEN mutations, shown or predicted to result in loss of function or dysfunction, and disease severity in PHTS have been suggested but will require further study (16, 18).

There is growing interest in understanding how various monoallelic inactivating tumor suppressor alterations lead to tumorigenesis. Experimental data from animal models suggest that lower Pten protein dosage influences carcinogenesis, with a Pten dose reduction of as little as 20% leading to murine mammary cancer development (19). Others have shown that homozygous deletion of Pten in mouse thyroid cells led to benign follicular adenomas in female mice (20), whereas hemizygous deletion of Pten was seen to accelerate thyroid adenocarcinoma formation when additional genetic alterations were present, such as loss of p27 (21) and in the setting of mutant thyroid hormone receptor-β (20). Based on these mouse studies, it does appear that decreased gene dosage of PTEN is indeed important for thyroid cancer progression in the presence of other genetic alterations. We demonstrated that, on the whole, reduced PTEN protein dose in CS lymphoblast-derived cell lines tended to occur in the presence of an underlying germline PTEN mutation and tended to correlate with increasing clinical phenotypic load of CS (17). Therefore, we sought to determine whether lowered levels of PTEN protein is predictive of an underlying germline PTEN mutation among CS and CS-like individuals who presented with thyroid cancer and, if so, whether loss of PTEN protein expression in affected malignant thyroid tissue can aid in the identification of patients in whom germline PTEN testing is warranted.

Materials and Methods

A total of 2792 research participants were prospectively accrued into and provided informed consent for Cleveland Clinic Institutional Review Board protocol 8458. Probands who met at least the relaxed International Cowden Consortium operational criteria for CS were eligible. Relaxed criteria are defined as full criteria minus one criterion, and such individuals are referred to as CS-like. These patients were recruited from both community and academic medical centers throughout North America, Europe, and Asia. For each patient, the medical record was examined by cancer genetics professionals, and when possible, primary documentation of medical records/pathology reports were obtained for confirmation of the thyroid cancer and precise histology, with the patients' consent.

PTEN mutation and deletion analysis

All research participants underwent PTEN (NM_000314.4) mutation analysis as described as follows. Genomic DNA was extracted from peripheral blood leukocytes using standard methods (22). Scanning of genomic DNA samples for PTEN mutations was performed as previously reported with a combination of denaturing gradient gel electrophoresis, high-resolution melting curve analysis (Idaho Technology, Salt Lake City, UT) and direct Sanger sequencing (ABI 3730xl; Life Technologies, Carlsbad, CA) (23). Deletion analysis using the multiplex ligation-dependent probe amplification (MLPA) assay (24) was performed with the P158 MLPA kit (MRC-Holland, Amsterdam, The Netherlands) according to the manufacturer's protocol. All patients underwent PCR-based Sanger sequencing of the PTEN promoter region (Supplemental Table 2 for primer sequences) as previously described (25). Promoter mutations were defined as previously reported (25, 26) except for −1084T→C. This variant has been reported in population controls of European descent (two in 150) (27), and additional work is required to characterize it. We consider it as a variant of unknown clinical significance (VUS) at this time.

Analysis of PTEN and other downstream proteins by immunoblotting

Human immortalized lymphoblast-derived cell lines were obtained from each patient and cultured in RPMI 1640 supplemented with 20% fetal bovine serum. All cell lines were cultured at 37 C and 5% CO₂. Whole-cell lysates were prepared with mammalian protein extraction reagent (ThermoFisher Scientific, Waltham, MA) supplemented with protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO). Lysates were separated by SDS-PAGE and transferred onto nitrocellulose. Antibodies used included anti-PTEN mouse monoclonal (Cascade Biosciences, Portland, OR) at 1:5000, anti-phospho-AKT rabbit polyclonal (Cell Signaling, Danvers, MA) at 1:1000, anti-phospho-MAPK1/2 rabbit polyclonal (Cell Signaling) at 1:20,000, anti-glyceraldehyde-3-phosphate dehydrogenase rabbit monoclonal (Cell Signaling) at 1:20,000, and anti-actin mouse monoclonal (Santa Cruz Biotechnology, Santa Cruz, CA) at 1:20,000. The blots were scanned digitally with the Odyssey imaging system (Li-Cor Biotechnology, Lincoln, NE). Detected fluorescence intensities for protein bands were background adjusted and normalized between gels with the median expression of individual proteins on each blot. Blood PTEN protein levels were classified as PTEN^Q1, PTEN^Q2, PTEN^Q3, and PTEN^Q4 with PTEN^Q1 corresponding to the lowest quartile of PTEN expression seen and PTEN^Q4 being the highest.

Immunohistochemical staining of thyroid tissue for PTEN

Four-micrometer-thick formalin-fixed paraffin-embedded tissue sections were cut, and immunohistochemistry was performed after pressure cooker antigen retrieval [0.01 m citrate buffer (pH 6.4) for 20 min]. To block endogenous peroxidase activity, the sections were incubated with 0.3% hydrogen peroxide in methanol for 30 min after cooling to room temperature. After blocking for 30 min in 0.75% normal horse serum, the sections were incubated with a mouse anti-PTEN monoclonal antibody (1:50 dilution; Cascade Biosciences, Portland, OR). Primary antibody binding was localized by using an avidin-biotin-peroxidase kit (Vector Laboratories, Burlingame, CA) according to the manufacturer's instruction. PTEN expression was independently evaluated by two of the authors (J.N. and X.H.) under a light microscope and randomly spot evaluated by a third investigator (C.E.). Both authors were blinded to the clinical history and underlying germline mutation. Intensity of PTEN staining was classified separately for the nucleus and the cytoplasm and graded strong (3+), moderate (2+), weak (1+), or absent (−).

Statistical analysis

The χ² test was used to measure the association between germline PTEN mutation status and categorical clinicopathological variables. Univariate and multivariate logistic regression models were used to test for the association between PTEN expression status, clinicopathological features, and germline PTEN mutation status. Odds ratios and 95% confidence intervals (CIs) were calculated. Likelihood ratio tests were used to evaluate the significance of each covariate. Performance of PTEN^Q1 in predicting PTEN mutation status was evaluated using receiver-operating characteristic (ROC) methodology to give measurements of sensitivity, specificity, accuracy, and positive and negative predictive values. Calibration to obtain bias-corrected estimates of predicted vs. observed values was performed with 500 bootstraps. Statistical tests were two sided with P ≤ 0.05 considered significant. All analyses were performed using SPSS version 19.0 software (SPSS Inc., Chicago, IL) and on R version 2.11.1 (28).

Results

Patient characteristics by germline PTEN status and blood PTEN protein category

Of the 2792 CS/CS-like patients, 721 had epithelial thyroid cancer confirmed by medical and/or pathology records. Germline PTEN status and blood PTEN protein levels were determined in all patients. Germline PTEN pathogenic mutations were found in 35 of 721 (5%) CS/CS-like patients (PTEN^mut+) with thyroid cancer. Human immortalized lymphoblast-derived cell lines were obtained from 582 patients (81%), and PTEN protein analyses were performed. Of 27 PTEN^mut+ patients who had PTEN protein analyses performed, 26 (96%) had blood PTEN expression levels in the lowest quartile (PTEN^Q1, Table 1), compared with 139 of 555 (25%) patients who did not harbor a germline PTEN mutation (P < 0.001). Age of thyroid cancer onset and presence of benign thyroid adenomas or thyroiditis was not associated with significantly different blood PTEN protein expression levels (Table 1), whereas male thyroid cancer and follicular histology subtype were more likely to be associated with a low blood PTEN^Q1 expression level (P = 0.05) (Table 1).

Table 1.

Relationship between clinicopathological characteristics of Cowden-related thyroid cancer patients, PTEN mutation status, and blood PTEN expression levels

Characteristics	No. of patients (n = 582)	Blood PTEN expression levels				P
		PTENQ1	PTENQ2	PTENQ3	PTENQ4
Age of thyroid cancer onset						0.29
≤18 yr	15	6	4	2	3
>18 yr	567	155	140	132	140
Sex						0.05
Male	35	16	10	5	5
Female	547	149	134	124	140
Histology						0.05
Follicular	56	19	7	10	20
Nonfollicular	526	146	137	119	124
Benign thyroid disease
Thyroiditis	84	30	22	14	18	0.29
Follicular adenoma/goiter	201	60	52	38	52	0.59
Germline PTEN mutation status						<0.001
Pathogenic mutation	27	26	0	0	1
VUS/SNP/WT	555	139	144	129	143

SNP, Single-nucleotide polymorphism; WT, wild type.

Multivariate logistic regression analysis of the 582 patients with blood PTEN protein expression was performed using the clinical features that were found to significantly predict for germline PTEN mutation by univariate analysis: age of thyroid cancer onset below 18 yr, male, presence of macrocephaly, follicular thyroid cancer histology and presence of coexisting thyroiditis, and/or thyroid adenoma and blood PTEN^Q1 expression. We found that five of the seven factors were independent predictors of the presence of a germline PTEN mutation (Table 2). Patients with blood PTEN^Q1 levels had a 73-fold increased risk of having an underlying germline PTEN mutation (P < 0.001). Presence of coexisting thyroid adenoma and a follicular thyroid cancer subtype were not found to be independent predictors in this model.

Table 2.

Multivariate analysis of predictive factors for pathogenic germline PTEN mutations in CS patients presenting with thyroid cancer

Factor	OR	95% CI	P
Male vs. female	4.96	1.35–18.18	0.02
Age of onset ≤18 vs. >18 yr	15.94	1.63–155.35	0.02
Macrocephaly vs. no macrocephaly	5.24	1.84–14.98	0.02
History of thyroiditis vs. no thyroiditis	3.31	1.04–10.57	0.04
PTENQ1 vs. PTENQ2,Q3,orQ4	73.51	9.27–583.17	<0.001

Relationship between type of pathogenic germline PTEN mutation and blood PTEN protein levels

PTEN frameshift and nonsense mutations accounted for 63% (17 of 27) of mutations. Of the 27 PTEN^mut+patients who had blood PTEN protein analyses performed, all patients with frameshift mutations (nine of nine), splice-site (two of two), missense (three of three), or nonsense mutations (eight of eight) had blood PTEN protein levels in the lowest quartile. Only one patient with a promoter mutation had protein levels that were higher than blood PTEN^Q1 (Fig. 1).

Fig. 1.

Mean PTEN fluorescent intensity according to PTEN mutation status and type of mutation. Error bars are representative of 95% CI. WT, Wild type.

Performance of blood PTEN^Q1 in predicting for germline PTEN mutations

A ROC curve was constructed to assess the ability of blood PTEN^Q1 to predict germline PTEN mutations. The area under the curve was 0.79 (95% CI = 0.68–0.88; Fig. 2). Low blood PTEN levels predicted for PTEN^mut+ cases with a sensitivity of 96.30% (95% CI = 80.97–99.38%), specificity of 74.95% (95% CI = 71.13–78.51%). Blood PTEN^Q1 gave a positive predictive value of 15.76% (95% CI = 10.56–22.23%) and a 99.76% negative predictive value (95% CI = 98.67–99.96%) and a positive test likelihood ratio of 3.84 (95% CI = 3.27–4.52). We used bootstrap validation to confirm that blood PTEN^Q1 within the dataset predicted for PTEN^mut+ cases (n = 500), demonstrating excellent calibration (Fig. 3).

Fig. 2.

ROC curve of low blood PTEN levels (PTEN^Q1) in predicting germline PTEN mutation in CS patients who present with thyroid cancer. AUC, Area under the curve.

Fig. 3.

Calibration plot for PTEN^Q1 predicted probabilities of PTEN mutation with actual outcomes.

Correlation between blood PTEN^Q1 and PTEN expression in affected thyroid tissue

Affected thyroid cancer tissues (n = 12) underwent PTEN immunohistochemical staining; seven from PTEN^mut+ cases showed global loss of PTEN protein by immunohistochemistry. Absent or weak PTEN expression on immunohistochemistry were seen in all cases and correlated strongly with patient-derived blood PTEN^Q1 protein levels (Fig. 4, A and B). Five cases of PTEN^wt/vus-related thyroid cancer showed strong cytoplasmic or nuclear PTEN staining (Fig. 4, C and D).

Fig. 4.

PTEN immunohistochemistry of germline PTEN^mut+-related thyroid cancer showing both nuclear and cytoplasmic loss of PTEN distinct from that seen in PTEN^wt/vus-related thyroid cancer (×40 magnification). A, Papillary thyroid cancer with absent (−) PTEN staining; underlying c.210–1G→A germline PTEN mutation with blood PTEN^Q1 expression; B, follicular variant of papillary thyroid cancer with weak (+) nuclear PTEN staining; underlying c.80-?_164+?del germline PTEN mutation with blood PTEN^Q1 expression; C, papillary thyroid cancer with strong (3+) cytoplasmic PTEN staining; PTEN wild-type with blood-PTEN^Q3 expression; D, papillary thyroid cancer with moderate (2+) nuclear PTEN staining; PTEN wild-type with blood-PTEN^Q1 expression.

Discussion

Germline mutations in PTEN support a haploinsufficient tumor suppressor model given that heterozygosity in CS patients lead to clinical phenotype and increased cancer susceptibility even in the absence of detectable biallelic mutation of the PTEN gene in tumors derived from CS patients (29, 30). Small changes in the expression levels of tumor suppressor genes have been proposed to influence susceptibility to cancer (19, 31). Our findings suggest that the functional PTEN dose is important. Our results show that in the setting of epithelial thyroid cancer presentation, changes resulting in low blood PTEN expression level was seen in 96% of PTEN^mut+ patients (26 of 27) and could be used to accurately exclude the absence of a germline PTEN mutation. In particular, it predicted for 100% of patients with frameshift, splice-site, large-deletion, missense, and nonsense mutations (Fig. 1).

All three patients with missense PTEN mutations had lower blood PTEN expression compared with patients who were PTEN^wt/vus. This makes sense given that all three of these mutations are located either close to the active site pocket of the phosphatase domain (C136R and S170I) or at the C-terminal region of PTEN (K254T) and are predicted to be pathogenic, consistent with in vitro functional studies (32, 33) showing mutations in these regions resulting in destabilization and decreased expression level. Therefore, it is not surprising to see low blood PTEN expression levels in patients harboring these germline alterations.

For clinicians, importantly, for the first time, we now have clinical predictors for underlying germline PTEN mutation in CS/CS-like patients presenting with thyroid cancer. Of all the factors examined, only four clinical factors were significant in independently predicting a germline PTEN mutation after multivariate logistic regression: presence of macrocephaly, male thyroid cancer, history of thyroiditis, and onset of thyroid cancer under the age of 18 yr, the presence of any of which should alert the clinician to the possibility of a germline PTEN mutation in CS/CS-like patients presenting with thyroid cancer.

Notably, blood PTEN^Q1 level was a stronger predictor than any clinical features. Impressively, a low blood PTEN expression level missed only one case with a germline PTEN mutation with a negative predictive value in excess of 99%. This suggests that perhaps blood PTEN protein expression levels could serve as a reliable molecular correlate for selecting patients in whom genetic testing may not be warranted. The likelihood of someone with a low blood PTEN^Q1 expression having an underlying germline PTEN mutation is 3.84-fold greater than in those who had higher blood PTEN expression levels. However, one limitation of this study is the absence of a normal range for blood PTEN expression levels. In our study, we were able to quantify the blood PTEN expression of the entire cohort of 721 CS and CS-like patients presenting with thyroid cancer and classified them into quartiles of blood PTEN expression for comparison between patients who had an underlying germline mutation and those who did not. Although this has performed well in this study, our data are preliminary and validation in unselected populations in the future should lead to direct clinical application.

Among patients who did not harbor any germline PTEN mutation, 25% of them had blood PTEN^Q1 levels. Of the 139 patients, two had a PTEN VUS, 32 with common PTEN single-nucleotide polymorphisms and the remaining 105 were PTEN wild-type. This affected both the positive predictive value and specificity of using blood PTEN^Q1 in predicting for patients with germline PTEN mutations. In the absence of germline mutations, other mechanisms that may lead to progressive loss of PTEN expression include epigenetic events and posttranslational modifications on PTEN, misguided PTEN subcellular localization, and PTEN-specific microRNA up-regulation (34–39). Blood PTEN^Q1 expression could be used to better characterize patients meeting criteria for CS/CS-like who do not harbor pathogenic germline PTEN mutations to help drive more research into other mechanisms by which PTEN is negatively regulated in these patients.

We saw a close correlation between low blood PTEN levels and weak/absent PTEN expression in affected thyroid tissue of PTEN^mut+ patients. We and others have shown that PTEN is strongly expressed in normal thyroid epithelium, in particular in the nucleus and to a lesser extent in the cytoplasm (13, 40, 41). Barletta et al. (42) recently showed that eight of nine patients who had low PTEN expression in a thyroidectomy series had clinical features suggestive of CS. Consistent with these findings, we saw globally (nuclear and cytoplasmic) reduced PTEN expression in affected thyroid adenomas and thyroid cancers. Although we were limited by the availability of affected PTEN^mut+ thyroid cancer tissue in this study, collectively, our and others' data suggest that thyroid tissue PTEN expression such as blood PTEN expression could serve as a clinically useful molecular correlate for germline PTEN mutation and warrants to be studied prospectively to see whether it can complement or surpasses current clinical predictors.

The classic two-hit model of tumor suppressor inactivation was originally established by mathematical modeling of cancer incidence and had implied that tumorigenesis requires complete loss of function of tumor-suppressor genes. Increasingly, we are beginning to see that the nuanced functional importance of PTEN dosage affects PTEN in its role as a tumor suppressor. We provide here both patient-derived support for preclinical data that PTEN protein dose may indeed influence tumorigenesis and a potential role whereby PTEN protein expression in thyroid tissue/blood can serve as a molecular correlate for selecting patients in whom genetic assessment and testing may be warranted.