TERT Promoter Mutations Occur Early in Urothelial Neoplasia and Are Biomarkers of Early Disease and Disease Recurrence in Urine

Urothelial carcinoma of the bladder is the most common malignancy of the urinary tract with 73,000 new cases and 15,000 deaths expected in 2013 in the United States alone (1). These invasive carcinomas arise from histologically well-defined papillary and flat precursor lesions, providing a potential opportunity for early detection and treatment (2). Although urine cytology enjoys a reasonable sensitivity and specificity for detecting high-grade neoplasms, its performance in detecting low-grade tumors is poor, with a sensitivity and specificity of 4% and 30%, respectively (3).

A number of urine-based markers have been developed to improve the accuracy of noninvasive screening and surveillance in bladder cancer. Among U.S. Food and Drug Administration–approved tests, the immunocy test (Scimedx Corp), nuclear matrix protein 22 (NMP22) immunoassay test (Matritech), and multitarget FISH (UroVysion; ref. 4) have demonstrated an overall sensitivity of 70% and a specificity range of up to 89%. Performance inconsistencies, as a result of variability in preanalytical and analytical specimen factors, have impeded their widespread clinical use.

Activating mutations in the promoter of the telomerase reverse transcriptase (TERT) gene lead to increased telomerase expression and, in doing so, allow some neoplasms to overcome the end-replication problem and avoid senescence. TERT promoter mutations were initially described in melanoma (5, 6) and have subsequently been described in a discrete spectrum of cancer types, including 66% of muscle-invasive urothelial carcinomas of the bladder (5, 7). TERT is therefore the most frequently mutated gene in advanced forms of this disease, and the localization of these mutations to a small gene region in the TERT promoter provides an extraordinary opportunity for biomarker development (7).

For TERT promoter mutations to be a useful marker of early, curable disease, these mutations should be present in preinvasive bladder tumors and shed into the urine. To this end, we have in this study evaluated the sequence of the TERT promoter in a large number of curable precursor neoplasms of the urinary bladder. We also determined the sequence of the TERT promoter in a separate group of superficial bladder cancers and corresponding follow-up urine samples to establish the feasibility of detecting TERT mutations in urine and their potential utility in predicting recurrence.

This study was approved by the Institutional Review Board of Johns Hopkins University, School of Medicine (Baltimore, MD). Two different sets of samples were analyzed in our study. The first sample set included 76 noninvasive papillary urothelial carcinomas and flat carcinoma in situ (CIS) lesions obtained by transurethral bladder resection (TURB) between 2000 and 2012. All specimens were from the Surgical Pathology archives and were selected only on the basis of specimen availability. Pertinent patient demographics and clinical information were obtained from electronic medical records. All sections were reviewed by three urological pathologists (E. Munari, S.F. Faraj, and G.J. Netto) to confirm the original diagnoses. To enrich for neoplastic cells within the tissues, representative formalin-fixed paraffin-embedded (FFPE) blocks were cored with a sterile 16-gauge needle, and tumor areas showing at least 50% neoplastic cellularity were selected microscopically. For eight of the cases, benign adjacent urothelium was macrodissected from FFPE blocks. The cores were placed in a 1.5-mL sterile tube for subsequent DNA purification using an AllPrep DNA/RNA Mini Kit (Qiagen, cat. no. 80204). DNA was purified from peripheral blood buffy coats of 15 patients using the same Qiagen kit.

For the second sample set, we prospectively collected urine samples from 15 separate patients undergoing follow-up cystoscopy for previously diagnosed nonmuscle-invasive urothelial carcinoma. We purposely biased this cohort to include patients that recurred within the follow-up period. Immediately prior to follow-up cystoscopy, 25 mL of raw urine was collected and subsequently pelleted by centrifugation at 3,000 g for 10 minutes. The pellets were stored at −80°C in 1.5 mL tubes for subsequent DNA extraction. For 14 of these patients, matched FFPE from the original diagnostic TURB was retrieved. These included 13 high-grade urothelial carcinomas (pTa HG and pT1 HG in six and seven cases, respectively), and one low-grade papillary urothelial carcinoma (pTa LG). Twenty-eight micrometer thick sections were cut from one representative tissue block in each case and areas containing at least 70% neoplastic cells were microdissected and used for DNA purification using a QIAamp DNA FFPE Tissue Kit (Qiagen, cat. no. 56404).

Because of their tremendous throughput, massively parallel sequencing instruments are highly cost effective for DNA mutation analysis. However, sample preparation and sequencing steps introduce artifactual mutations into analyses at a low, but significant frequency. To better discriminate genuine TERT promoter mutations from artifactual sequencing variants introduced during the sequencing process, we used Safe-SeqS, a sequencing error-reduction technology described previously (8, 9). As depicted in Fig. 1, Safe-SeqS amplification primers were designed to amplify a 126-bp or 76-bp segment containing the region of the TERT promoter previously shown to harbor mutations in melanomas and other tumors (5–7). The forward and reverse amplification primers contained the TERT-specific sequences at their 3′ ends and a universal priming site (UPS) at their 5′ end. The reverse primer additionally contained a 14-base unique identifier (UID), composed of 14 degenerate N bases (equal likelihood of being an A, C, T, or G) between the UPS and gene-specific sequences. The sequences of the forward and reverse primers were either 5′-CACACAGGAAACAGCTATGACCATGGGCCGCGGAAAGGGAAG and 5′-CGACGTAAAACGACGGCCAGTNNNNNNNNNNNNNNCGTCCTGCCCCTTCACC, or 5′-CACACAGGAAACAGCTATGACCATGGCGGAAAGGAAAGGGAG and 5′-CGACGTAAAACGACGGCCAGTNNNNNNNNNNNNNNCCCGTCCCGACCCCT (UPS sequences underlined). These primers were used to amplify DNA in 25-μL PCR reactions in 1X Phusion Flash High-Fidelity PCR Master Mix (Thermo Scientific, cat. no. F-548L) containing 0.5 μmol/L forward and reverse primers (described above). After incubation at 98°C for 120 seconds, 10 cycles of PCR were performed in the following manner: 98°C for 10 seconds, 63°C for 120 seconds, and 72°C for 120 seconds. Reactions were purified with AMPure XP beads (Beckman Coulter) and eluted in 100 μL of Buffer EB (Qiagen, cat. no. 19086). For the second stage of amplification, 5 μL of purified PCR products were amplified in 25 μL reactions containing 1X Phusion Flash High-Fidelity PCR Master Mix and 0.5 μmol/L amplification primers that each contained the first-stage UPS at their 3′ ends and the grafting sequences required to hybridize to the sequencing instrument flow cell at their 5′ ends (8, 9). The reverse amplification primer additionally included a 6 bp index sequence, unique to each sample, inserted between the UPS and grafting sequences. After incubation at 98°C for 120 seconds, 17 cycles of PCR were performed in the following manner: 98°C for 10 seconds, 63°C for 120 seconds, and 72°C for 120 seconds. The PCR products were purified with AMPure and sequenced on a MiSeq instrument.

Data were analyzed as previously described (8, 9). Briefly, the amplified TERT promoter region of reads containing UIDs, where each base of the UID region had instrument-derived quality scores more than or equal to 15, was matched to a reference sequence using a custom script. TERT promoter sequences with five or fewer mismatches were retained for further analysis. Tumor samples were considered positive if the fraction of mutations exceeded 1% of alleles (which was a frequency at least 10× higher than that found in control DNA templates from FFPE tissues). Urine samples were considered positive when the frequency of mutation exceeded 0.1% of alleles (a frequency at least 10× higher than that found in control DNA templates from urine samples of patients without TERT mutations in their primary tumors). All sequencing assays scored as positive were confirmed in at least one additional, independent PCR and sequence assay.

The data were analyzed using Stata/SE 12 (StataCorp Inc.). A Pearson χ² test was used for the analysis of the association of categorical variables. A two-tailed probability less than 0.05 was required for statistical significance.

We used a massively parallel sequencing technology to determine the presence and representation of mutant TERT promoter alleles in urothelial cancers. A graphical depiction of the method is shown in Fig. 1 and detailed procedures are provided in Materials and Methods. In addition to revealing whether mutations are present with a population of DNA templates, this technique provides an accurate determination of the fraction of mutant alleles in the sample. Clinicopathologic characteristics of the 76 noninvasive urothelial carcinomas analyzed in the first phase of this study are summarized in Table 1. They included 59 papillary tumors—28 low-grade (pTa LG) and 31 high-grade (pTa HG)—plus 17 “flat” urothelial CIS. These patients were typical of those with this form of cancer; their average age was 66 years and most (82%) were males (Table 1).

TERT promoter mutations were identified in 56 of 76 (74%) of these urothelial carcinomas (Table 2). In contrast, none of the eight samples of adjacent normal urothelium harbored TERT promoter mutations. In addition, we did not detect TERT promoter mutations in 15 samples of peripheral blood from the same patients. Twelve of the blood samples and five of the normal urothelial samples were from patients whose tumors harbored TERT promoter mutations. These data demonstrate that the TERT promoter mutations in these patients were unequivocally somatic and limited to the neoplastic urothelium in the bladder. The predominant alterations were g.1295228C>T (minus strand of chromosome 5, hg19 assembly) and g.1295250C>T mutations, which accounted for 75% and 20% of the total alterations, respectively. In addition, we identified one g.1295228C>A mutation and two g.1295242C>T mutations not previously reported (Supplementary Table S1). The mutations were found in all types and grades of these early cancers: in 76% of papillary lesions and 65% of flat lesions; in 86% of low-grade and in 68% of high-grade lesions (Table 2). None of these differences among subgroups were statistically significant.

The results described above show that TERT promoter mutations occur early in bladder cancers and did not correlate with grade or type. Such early mutations would not be likely associated with recurrence or progression, but to evaluate this possibility, our series of samples included cases both with and without recurrence during follow up. In Tables 3 and 4, the relationship between TERT promoter mutation status and tumor recurrence or progression, respectively, are displayed. TERT promoter mutation status was not associated with likelihood of recurrence or progression in any subgroup.

We next evaluated whether TERT promoter mutations could be identified in cells in the urine. As noted in Introduction, urine samples are routinely taken at follow-up visits following TURB procedures to help determine whether residual tumor cells are present (via cytology or other methods). We first assessed the tumors obtained from 14 patients undergoing TURB for relatively early (nonmuscle invasive) disease. Of these, 11 (79%) harbored TERT promoter mutations (Table 5), as expected from the evaluation of the first cohort (Table 2). All of the mutations in the second cohort were at either g.1295228C>T or g.1295250C>T (Table 5).

The 14 patients were monitored for recurrence at subsequent visits. Mutations were assessed in the cell pellets from the urines obtained at the first follow-up visit after TURB in these 14 patients, as described in Materials and Methods. There was a striking correlation between the presence of a TERT promoter mutation in the urine, the presence of the mutation in the original tumor, and recurrence. In three of 14 patients without a TERT promoter mutation in their tumor, no mutation was evident in their urine sample, as expected (Table 5). Of the 11 patients in whom a TERT mutation was present in the tumor, seven patients were observed to have a mutation in the DNA isolated from their urine cell pellets; in each case, the mutation was identical to that observed in the primary tumor removed via TURB (Table 5). The bladder cancers in each of these seven patients recurred, either at the first follow-up or thereafter. The proportion of mutant alleles in the cells pelleted from the urine of these patients was often substantial, ranging from 0.17% to 23% with a median of 4.4% (Table 5). We also identified a TERT promoter mutation in a urine sample from which no prior tumor was available; this tumor also recurred (Table 5). In contrast, no TERT mutations were evident in the urine samples of four patients whose original tumors contained a TERT promoter mutation: the tumors of three of these patients never recurred, whereas the fourth developed a recurrence 3.5 months after the urine sample was collected (Table 5). As shown in Table 6, the presence of detectable TERT promoter mutations in the urine was strongly associated with recurrence of urothelial carcinoma (P < 0.001; Pearson correlation coefficient = 0.87).

TERT promoter mutations are detectable in urine, and their presence in urine is strongly associated with bladder cancer recurrence. Muscle-invasive urothelial carcinoma is responsible for the vast majority of deaths related to bladder cancer and many of these deaths could be prevented if precursor lesions were detected and surgically excised prior to their invasion into the muscle (10–13). New strategies for the early detection of such lesions are therefore urgently needed (14). Our results show that TERT promoter mutations are the most common genetic alteration in noninvasive bladder cancer identified to date, occurring in the majority (74%) of such precursor lesions. They occur in cancers developing through both the papillary and flat routes to tumor progression (15), and occur in low-grade as well as high-grade tumors. We also show that these mutations can be detected in the urine of patients with bladder cancer. Altogether, these results suggest that TERT promoter mutations may provide a useful biomarker for the early detection of bladder cancers in the future, and that prospective studies of patients at high risk for this disease are warranted.

Given the high prevalence of TERT promoter mutations in early bladder neoplasia, their presence or absence in tumors is of limited prognostic value. However, superficial bladder cancers are currently the most costly solid tumor (per patient) in the United States (16, 17). Noninvasive methods to monitor these patients could reduce the cost of caring for these patients as well as the discomfort associated with invasive procedures. Our results are highly encouraging with respect to this potential application. Among patients with TERT mutations in their primary tumors, there was a highly significant correlation between the presence of mutations in subsequent urine collections and recurrence (Table 6).

Our results therefore suggest two potential avenues for application of TERT promoter mutations in the clinic: early detection in high-risk patients and monitoring of patients with bladder cancer, both through the analysis of urine specimens. It is important to note that both these applications will require further study prior to implementation. For example, we have not yet shown that patients with bladder cancer have detectable mutations in urine prior to tumor diagnosis; all of our urine samples were taken at follow-up visits after surgery. In addition, our study involved only a small number of patients, and we are yet to demonstrate that the analysis of urine for TERT mutations improves upon conventional cytology or clinical criteria, or whether it could partially replace cystoscopy in certain circumstances. Still, our study provides a strong proof of principle: TERT promoter mutations occur early, are specific for neoplasia, and can be identified in the urine with currently available technologies. Future large-scale studies will be required to determine the clinical utility of this approach for screening or monitoring purposes.

M.P. Schoenberg is employed as a consultant in SonaCare Medical, Tengion, Heat Bio, Photocure, and TheraCoat, and has ownership interest (including patents) in SonaCare Medical. K.W. Kinzler, L.A. Diaz, N. Papadopoulos, and B.Vogelstein are founders of Personal Genome Diagnostics, Inc., a company focused on the identification of genetic alterations in human cancer for diagnostic and therapeutic purposes. K.W. Kinzler and B. Vogelstein are also members of the Consultants of Sysmex-Inostics, a company that is developing technologies for the molecular diagnosis of cancer. These companies and others have licensed patent applications from Johns Hopkins relevant to the current study. The terms of these arrangements are being managed by the university in accordance with its conflict of interest policies. No potential conflicts of interest were disclosed by the other authors.

Conception and design: R.H. Hruban, M. Schoenberg, K.W. Kinzler, B. Vogelstein, L.A. Diaz, N. Papadopoulos, G.J. Netto, M.E. Allaf

Development of methodology: I. Kinde, E. Murani, L.A. Diaz, K.W. Kinzler, B. Vogelstein, N. Papadopoulos, G.J. Netto

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): I. Kinde, S.F. Faraj, M.P. Schoenberg, T. Bivalacqua, M.E. Allaf, S. Springer, Y. Wang, L.A. Diaz, K.W. Kinzler, B. Vogelstein, N. Papadopoulos, G.J. Netto

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): I. Kinde, E. Munari, S.F. Faraj, R.H. Hruban, M. Schoenberg, L.A. Diaz, K.W. Kinzler, B. Vogelstein, N. Papadopoulos, G.J. Netto

Writing, review, and/or revision of the manuscript: I. Kinde, E. Munari, S.F. Faraj, R.H. Hruban, M. Schoenberg, T. Bivalacqua, L.A. Diaz, K.W. Kinzler, B. Vogelstein, N. Papadopoulos, G.J. Netto, M.E. Allaf

The authors thank J. Schaeffer, J. Ptak, N. Silliman, M. Whalen, L. Dobbyn, and T. Driscoll for expert technical assistance.

This work was supported by the Lustgarten Foundation for Pancreatic Cancer Research; the Virginia and D. K. Ludwig Fund for Cancer Research; The Sol Goldman Pancreatic Cancer Research Center; the Joseph L. Rabinowitz Fund; the Michael Rolfe Foundation; the Stringer Foundation; the Brady Urological Institute (Johns Hopkins) Patana Fund; the Flight Attendant Medical Research Institute (Miami, FL) Clinical Innovator Award; the family of Ted and Julie Smith; the Indiana Genomics Initiative of Indiana University, which is supported, in part, by Lilly Endowment, Inc.; the J. C. Monastra Foundation; Swim Across America; National Science Foundation Grant DBI-1845275; NIH grants CA 62924, CA 57345, CA 152432, and CA 43460; and National Cancer Institute Contracts N01-CN-43302, N01-CN-43309, and NCI grants PO1CA077664 and P01CA134292.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.