TERT promoter mutations and Ki-67 labeling index as a prognostic marker of papillary thyroid carcinomas: combination of two independent factors

Although most papillary thyroid carcinomas (PTCs) have a good prognosis, a small but certain fraction shows aggressive behavior. Therefore, a novel and well-performing molecular marker is needed. In the present study, we assessed the impact of the combination of the TERT promoter/BRAF mutations and Ki-67 labeling index (LI) as a prognostic marker in PTC patients. Of 400 PTC samples, 354 were successfully genotyped for both TERT promoter/BRAF and analyzed for Ki-67 LI. Based on the combination of the mutational status and Ki-67 LI, the cases were categorized into three groups: high-, middle-, and low-risk. The recurrence rates of low-, middle-, and high-risk group were 1.9% (6 of 323), 18.2% (4 of 22), and 44.4% (4 of 9), respectively. The Kaplan-Meier curve and log-rank analyses demonstrated that there were statistical differences between any two groups. The hazard ratios for recurrence remained significant after adjustment for age, sex, tumor size, and extrathyroidal extension (low vs. middle: 8.80, 95% CI: 2.35–32.92, p = 0.001; middle vs. high: 6.255, 95% CI: 1.13–34.51, p = 0.035). In conclusion, the combination of the TERT promoter/BRAFV600E mutations and Ki-67 LI performed excellent in predicting PTC recurrence and may be clinically useful.

Relationship between the mutational status and clinicopathological features. We analyzed the association of BRAF V600E alone and coexistence of BRAF V600E and the TERT promoter mutation with clinicopathological characteristics. Unfortunately, in this series it was not possible to address the effect of the TERT promoter mutation alone due to absence of such cases. As shown in Table 1, the mean age was significantly older in the TERT promoter mutation-positive group (70.1 ± 8.3 years old) than in the BRAF V600E alone (49.7 ± 14.6 years old, p < 0.001) or mutation-negative group (45.8 ± 16.6 years old, p < 0.001). Interestingly, the TERT promoter mutation was not found at all in patients less than 45 years old, and its prevalence was substantially increased with age afterwards (Fig. 1). The tumor size was significantly greater in the TERT promoter mutation-positive group (32.1 ± 15.7 mm) than in the BRAF V600E alone (19.8 ± 11.4 mm, p < 0.001) or mutation-negative group (22.6 ± 15.3 y, p = 0.001) ( Table 1). Distant metastasis, advanced stage, and extrathyroidal invasion were also more common in the TERT promoter mutation-positive group while no difference was found for nodal disease frequency among the groups (please see Table 1 for details). These data demonstrate that the TERT promoter mutations were associated with the aggressive clinicopathological characteristics of PTC.    Relationship between the Ki-67 LI and recurrence. The recurrence rates of the LI < 5%, the LI 5-10%, and the LI> 10% groups were 2.0% (6 of 304), 4.4% (3 of 68), and 26.1% (6 of 23), respectively. Kaplan-Meier and log-rank analyses demonstrated that the LI> 10% group showed significantly worse recurrence-free survival (p < 0.001 vs LI < 5%, p = 0.003 vs LI 5-10%) (Fig. 2B). However, there was no statistical difference between LI < 5% and LI 5-10%. The HR for recurrence in patients with LI> 10% was 9.87 (95% CI: 3.42-28.5, p < 0.001), which remained significant after adjustment for age, sex, lymph node metastasis, tumor size, and extrathyroidal extension (HR: 5.521, 95% CI: 1.72-17.74, p = 0.004) ( Combination of the TERT promoter mutations and the Ki-67 LI. As described above, the associations of the TERT promoter mutations with the clinicopathological parameters were stronger as compared to those of the Ki-67 LI. Therefore, we combined both mutational status and LI to re-categorize the patients into three following groups: low-risk group, TERT mutation negative/LI ≤ 10% and TERT mutation positive/LI < 5%; middle-risk group, TERT mutation negative/LI> 10% and TERT mutation positive/LI 5-10%; high-risk group, TERT mutation positive/LI> 10% (Fig. 3A). The recurrence rates of low-, middle-, and high-risk group were 1.9% (6 of 323), 18.2% (4 of 22), and 44.4% (4 of 9), respectively (Fig. 3A). As shown in Fig 3B, the Kaplan-Meier curve of the low-risk group was excellent; that of the high-risk group was the worst; and that of the middle-risk group was in between. There were statistical differences between any two groups (Fig. 3B)

Discussion
The prevalence of the BRAF V600E mutation in PTCs in East Asian countries including Japan is high. The prevalence of the BRAF V600E mutation was reported to be dependent on the amount of iodine intake 29 ; however, very recently it has been demonstrated that there was no difference in the BRAF V600E prevalence between Japan, an iodine-rich country, and Vietnam, an iodine-deficient country 30 . Interestingly, in the meta-analysis by Kim et al. 31 , when they divided analyzed studies into two groups according to the prevalence of the BRAF V600E mutation (≥ 50% and < 50%), the pooled effect sizes of extrathyroidal invasion and lymph node metastasis were about 30% smaller in the BRAF V600E ≥ 50% group than in the < 50% group. The impact of the BRAF V600E mutation on aggressive clinicopathological features might depend on the prevalence of the mutation. In the present study, the association of the BRAF V600E mutation alone with the aggressive clinicopathological features and recurrence was negative. The reason for the difference in the impact of the BRAF V600E mutation remains to be further addressed. The rate of the TERT promoter mutations (C288T and C250T) in the current series was 10.1%, which is consistent with most of the previous studies 32, 33 . There seems to be no large difference between populations except for two studies from China reporting 4.1 and 4.4% 19,23 , perhaps the smallest two among published data. The presence of the TERT promoter mutation only was also reported to be associated with PTC aggressiveness 24 ; however, in the present study we were not able to analyze this because we did not have even one case with the TERT promoter mutation but without BRAF V600E , which is a limitation of our work. To assess the impact of the TERT promoter mutation alone in the BRAF V600E prevalent region such as Japan, it would be necessary to collect a very large number of samples. This is a future issue but cases with TERT promoter mutation alone probably show better prognosis than those with the TERT promoter plus BRAF V600E mutations. A number of studies have demonstrated that the co-existence of the BRAF V600E and TERT promoter mutations are associated with aggressive clinicopathological features and disease recurrence [17][18][19][20][21][22][23][24] . As far as we know, there is no report against the above finding so far. The result of the present study also supports this. We also observed that the presence of the TERT promoter mutations showed very strong age dependency, and it is noteworthy that the mutations were not   detected in any of the patients less than 45 years old. Nevertheless, the HR for recurrence after adjustment for age was still statistically significant. The TERT promoter mutations were linked to many aggressive clinicopathological features such as large tumor size, distant metastasis, advanced stage, and extrathyroidal extension. Therefore, it was not too surprising that the statistical significance of the increased HR for recurrence was lost after adjustment for tumor size or extrathyroidal extension in addition to age and sex. However, as their effect sizes were still relatively high (3.016 [adjusted for age, sex, and size], 4.068[adjusted for age, sex, Ex]), the number of samples was perhaps not enough to reach statistical significance. Our ROC curve analysis demonstrated that mutation type would not be a good test for predicting recurrence in the series under analysis (AUC = 0.672). This is a first study reporting a clinical significance of the BRAF V600E and TERT promoter mutations in the Japanese PTC cases, which show higher BRAF V600E prevalence (81.7% in this study). This is consistent with the recent report from South Korea where the prevalence of the BRAF V600E mutation is also high 20 .
It has been demonstrated that the Ki-67 LI is associated with a prognosis of PTC 26,27 . The current study also confirmed these observations. Indeed, the HR for recurrence of the Ki-67 LI> 10% was significant even after adjustment for a number of parameters, and effect size was greater than that for the TERT promoter/BRAF V600E mutations. Also, ROC curve analysis indicated that Ki-67 LI performed better than mutation type as a test for predicting recurrence (AUC = 0.727).
Interestingly, the associations of the Ki-67 LI with the clinicopathological characteristics were much weaker than those of the TERT promoter/BRAF V600E mutations, suggesting that Ki-67 LI may be a prognostic marker that is associated with higher-risk tumors independently of their mutational status. Indeed, the Ki-67 LI was reported to have an inverse correlation with the thyroglobulin-doubling time, suggesting that it reflects growth velocity of tumor 27 . On the other hand, the TERT promoter/BRAF V600E mutations are strongly linked to the clinicopathological status. They may be independent factors. We, therefore, combined the mutational status and the Ki-67 LI and divided the cases into three risk groups and found 6-8-fold adjusted differences in effect sizes between the groups. Concordantly, the ROC curve analysis based on the risk groups demonstrated further improvement of the test performance (AUC = 0.889). It should be noted that, despite the presence of the TERT promoter/BRAF V600E mutations, there was no recurrence in the patients with Ki-67 LI < 5% (low-risk group). In contrast, 16.7% of the PTCs recurred if Ki-67 LI was > 10% even though the tumors did not carry the TERT promoter/BRAF V600E mutations (middle-risk group). Approximately half of the high-risk patients had disease recurrence, although the number of such patients was rather small. Thus, neither mutational status nor Ki-67 LI was the optimal predictor for recurrence. Our statistical calculations showed that the combination of these risk factors performs better, yet ample room for improvement still remains.
In conclusion, the combination of the TERT promoter/BRAF V600E mutations and Ki-67 LI is a promising marker to predict recurrence of PTC. Low-, middle-, or high-risk group showed clear differences in any of the comparisons. The limitations of this method are: 1) The Ki-67 LI can only be obtained using surgical specimens, 2) It may be difficult to compare the Ki-67 LI from different institutions. Therefore, it is desired to find an objective marker which is strongly correlated with Ki-67 LI that can be measured preoperatively.

PTC samples.
A total of 400 adult sporadic PTC samples were collected at Kuma Hospital (Kobe, Japan). All patients received surgical treatment in 2009, and the therapeutic strategy was not changed during whole treatment course. Histological diagnosis was performed by a thyroid pathologist (MH). All patients had no history of radiation exposure. Patients' age at operation ranged 13-87 years old (mean 51 ± 16, median 52 years old, 15.2% male); follow-up period was 2-72 months (mean 58 ± 14, median 63 months). Disease recurrence was defined as surgically removed and pathologically verified local tumor focus or regional metastasis, or distant metastasis detected by ultrasound or radioisotope imaging not earlier than 12 months after initial treatment. According to this criterion, we excluded one case that had a recurrence within three months. The study protocol was approved by the ethics committees of Nagasaki University and Kuma Hospital. All procedures were performed in accordance with the relevant guidelines and regulations. DNA extraction and mutation screening. DNA was extracted from formalin-fixed paraffin-embedded (FFPE) PTC tissues using a QIAamp DNA mini kit (QIAGEN) according to the manufacturer's protocol. DNAs of sufficient quality and quantity for sequencing were obtained from 398 of 400 PTC specimens. BRAF (around V600) and TERT promoter mutations were analyzed by direct DNA sequencing. Primer sequences used for both PCR amplification and sequencing are: BRAF-FW, 5′ -ACATACTTATTGACTCTAAGAGGAAAGATGAA-3′ ; BRAF-RV, 5′ -GATTTTTGTGAATACTGGGAACTATGA-3′ ; TERT-FW, 5′ -CAGCGCTGCCTGAAACTC-3′ ; and TERT-RV, 5′ -GTCCTGCCCCTTCACCTT-3′ 22 . First, PCR amplification was done using KOD FX (TOYOBO). PCR products were then treated with ExoSAP-IT PCR clean-up reagent (GE Healthcare), and sequencing was performed with a Big Dye Terminator sequencing kit version 3.1 (Applied Biosystems) on an ABI3730 automated sequencer (Applied Biosystems). We prepared one negative control (without tissue section) per every 23 samples during DNA extraction to ensure contamination-free amplifications.
Ki-67 immunohistochemistry and the LI. Immunostaining was performed using 4-μ m-thick FFPE sections of the same PTC cases. Anti-Ki-67 antibody (clone MIB1, Dako) was used as a primary antibody. The staining was carried out using the Dako Cytomation Autostainer Universal System (Dako) and the Envision kit (Dako) according to the manufacturer's instruction. A single pathologist (MH) evaluated the specimens without clinical information of the patient. To obtain the Ki-67 LI, at least 500 carcinoma cells in hot areas were analyzed under × 400 magnification. Staining results were classified into three groups:< 5%, 5%-10%, and > 10% of positive cells.
Scientific RepoRts | 7:41752 | DOI: 10.1038/srep41752 Statistical analysis. Statistical analysis was performed using SPSS software version 21.0.0.0 (IBM), GraphPad Prism version 6.0 (GraphPad Software), and SAS University Edition software (SAS Institute Inc). For multiple comparisons, the nonparametric one-way ANOVA with Dunnett post hoc test or the FREQ and COMPPROP procedures (SAS) were used for continuous variables or group analyses, respectively. For univariate disease-free survival, the log-rank and the Kaplan-Meier estimates were calculated, and the Cox proportional hazard model was applied in multivariate analyses. The Receiver Operating Characteristic (ROC) curve analysis was performed under parametric distribution assumption (Eng J. ROC analysis: web-based calculator for ROC curves. Baltimore: Johns Hopkins University. Available at: http://www.jrocfit.org/, accessed on July 12, 2016). The p value less than 0.05 was regarded as indicating statistical significance.