Radiomics in predicting mutation status for thyroid cancer: A preliminary study using radiomics features for predicting BRAFV600E mutations in papillary thyroid carcinoma

Radiomics in predicting mutation status for thyroid cancer: A preliminary study using radiomics... OPENACCESS Citation: Yoon JH, Han K, Lee E, Lee J, Kim E-K, Moon HJ, et al. (2020) Radiomics in predicting Purpose mutation status for thyroid cancer: A preliminary To evaluate whether if ultrasonography (US)-based radiomics enables prediction of the study using radiomics features for predicting V600E V600E presence of BRAF mutations among patients diagnosed as papillary thyroid carcni- BRAF mutations in papillary thyroid carcinoma. PLoS ONE 15(2): e0228968. https:// noma (PTC). doi.org/10.1371/journal.pone.0228968 Editor: Jeeun Kang, Johns Hopkins University, Methods UNITED STATES From December 2015 to May 2017, 527 patients who had been treated surgically for PTC V600E Received: August 2, 2019 were included (training: 387, validation: 140). All patients had BRAF mutation analysis Accepted: January 27, 2020 performed on surgical specimen. Feature extraction was performed using preoperative US images of the 527 patients (mean size of PTC: 16.4mm±7.9, range, 10–85 mm). A Radio- Published: February 13, 2020 mics Score was generated by using the least absolute shrinkage and selection operator Copyright:© 2020 Yoon et al. This is an open (LASSO) regression model. Univariable/multivariable logistic regression analysis was per- access article distributed under the terms of the V600E formed to evaluate the factors including Radiomics Score in predicting BRAF mutation. Creative Commons Attribution License, which permits unrestricted use, distribution, and Subgroup analysis including conventional PTC<20-mm (n = 389) was performed (training: reproduction in any medium, provided the original 280, validation: 109). author and source are credited. Data Availability Statement: Data cannot be Results shared publicly because of patient confidentiality Of the 527 patients diagnosed with PTC, 428 (81.2%) were positive and 99 (18.8%) were issues. The institutional review board of Severance V600E Hospital, Yonsei University has given approval for negative for BRAF mutation. In both total 527 cancers and 389 conventional PTC<20- this study, recommending patient confidentiality mm, Radiomics Score was the single factor showing significant association to the presence when using image data. contact: hpc@yuhs.ac. V600E of BRAF mutation on multivariable analysis (all P<0.05). C-statistics for the validation Funding: This work was supported by the National set in the total cancers and the conventional PTCs<20-mm were lower than that of the train- Research Foundation of Korea(NRF) grant funded ing set: 0.629 (95% CI: 0.516–0.742) to 0.718 (95% CI: 0.650–0.786), and 0.567 (95% CI: by the Korea government(MSIT) (2019R1A2C1002375). This study was also 0.434–0.699) to 0.729 (95% CI: 0.632–0.826), respectively. PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 1 / 11 V600E Radiomics in predicting BRAF mutation supported by a CMB-Yuhan research grant of Conclusion Yonsei University College of Medicine (6-2017- Radiomics features extracted from US has limited value as a non-invasive biomarker for 0170). The funders had no role in study design, V600E data collection and analysis, decision to publish, or predicting the presence of BRAF mutation status of PTC regardless of size. preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. Introduction During the past decade, the incidence of thyroid cancer has rapidly increased worldwide, regardless of the demographic groups [1–3]. The majority of thyroid cancers that are being newly detected are papillary thyroid cancers (PTC) [4,5], a subtype consisting of more than 80% of all differentiated thyroid carcinomas [5,6]. In general, PTC is known to have excellent patient outcomes, 5-year survival rates approaching 98–99% [7,8], but approximately 10–15% of patients had aggressive tumor behavior, local recurrence/distant metastasis after treatment or mortality [9–12]. At present, even with the well-known poor prognostic factors such as age over 45 years, male gender, radioactive iodine resistance [9], it is difficult to descriminate to predict which patient has more aggressive forms of PTCs, and effort has been made using vari- ous biomarkers in predicting PTC patients with poor outcome. With the advancement in molecular genetics, various genetic alterations have been revealed and used as an adjunctive diagnostic method or for predicting patient prognosis [8,13,14]. V600E BRAF mutation, the most frequent oncogene in PTC, has been reported to be associated with aggressive clinical features such as large tumor size, extrathyroidal extension and presence of lymph node metastasis [8,14–16], leading to recurrence or mortality. But even with the abil- ity of either mutation in detecting aggressive cancer types, genetic analysis requires specimen tissue for analysis, mostly obtained from invasive surgical procedures. Aside from the informa- tion obtained from conventional imaging, radiomics, using data extracted from medical images converted into high-dimensional, mineable, and quantitative imaging features has been applied to revealing tumor physiology. Other studies have linked imaging features to molecular properties of tumors among various organs [17–20], but to the best of our knowl- edge, no studies have applied radiomics in predicting molecular status of thyroid cancer that can be used in predicting tumor aggressiveness. Based on this, we evaluated whether if ultraso- V600E nography (US)-based radiomics enables prediction of the presence of BRAF mutations among patients diagnosed as PTC. Materials & methods This retrospective study has been approved by the institutional review board (IRB) of Sever- ance Hospital, Yonsei University (approval number: 4-2018-0172), with a waiver for patient consent due to the retrospective study design. Signed informed consent was obtained from all patients prior to biopsy or surgical procedures. Images used for data extraction were fully anonymized before data processing according to the instructions of our IRB. Patients We included 527 patients who had been treated surgically with cytologically-proven or suspi- cious thyroid cancer between December 2015 to May 2017 at Severance Hospital, Seoul, V600E Korea. All patients had BRAF mutation analysis performed on surgical specimen. The 387 consecutive patients who had surgery from December 2015 to December 2016 were used as the training cohort: 300 women, 87 men, mean age, 42.1 years±14.0 (range, 15–82 years). PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 2 / 11 V600E Radiomics in predicting BRAF mutation The 140 consecutive patients who had surgery from January 2017 to May 2017 were used as the validation cohort: 105 women, 35 men, mean age, 41.3 years±13.4 (range, 15–74 years). Mean age of the 527 patients were 41.9 years±13.8 (range, 15–82 years). Mean size of the thy- roid masses was 16.4 mm±7.9 (range, 10–85 mm). V600E As the frequency of BRAF mutation has been reported to be associated with tumor size and conventional PTC [21], subgroup analysis was performed including thyroid cancers confirmed as conventional PTCs measuring <20-mm (n = 389). Mean age of the 389 patients was 42.9 years±13.1 (range, 15–80 years). Mean size of the conventional PTCs was 14.9 mm ±4.6 (range, 10–19 mm). Clinicopathologic data regarding tumor size, lymph node metastasis were obtained from review of medical records. Imaging features of the thyroid masses used for analysis were obtained from an institutional database. US image selection and feature extraction One radiologist (J.Y.K.) reviewed the preoperative US examinations of the 527 patients on the picture archiving and communication system (PACS) and selected representative transverse or longitudinal images of the tumor. The selected representative images were converted into JPEG files for manual segmentation. One radiologist (J.H.Y.) who had 9 years of experience in thyroid imaging manually set a region-of-interest (ROI) along the boundary of the selected tumor using Paint software of Windows (Fig 1). Since ROI marking with colored brush using Paint software alters original intensities in image, the manual ROI segmentation is conducted over the duplicate images of collected JPEG files. Before starting ROI extracting procedure, all images were normalized for fair comparison. First the location information of ROI marking (coordinate information of red curves in Fig 1) was sought and then applied to the original JPEG image to extract ROI only. This procedure ensures that the original intensity of the Fig 1. Representative image of tumor segmentation using thyroid US. A diagonal region-of-interest (ROI) was drawn along the tumor border (red line) for feature extraction. https://doi.org/10.1371/journal.pone.0228968.g001 PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 3 / 11 V600E Radiomics in predicting BRAF mutation image was not affected by the ROI extraction process. Once ROI was extracted, a total of 730 feature information were gathered. The 730 features include the first order statistics (energy, entropy, kurtosis, skewness and so on), the second order statistics (the gray level co-occurrence matrix (GLCM) and gray level run-length matrix (GLRLM) were established and the corre- sponding features were extracted), and features from four discrete one-level wavelet decompo- sitions. The detailed calculation for these features can be found in [22]. To obtain the feature quantities, the house code in MATLAB 2018b was used. Here, 256 bins using a bin with of 1 were utilized for intensity histogram and 4 angles of 0, 45, 90, and 135 degrees were utilized for GLCM and GLRLM anaylsis. V600E BRAF mutational analysis Direct DNA sequencing was used for the surgical specimen in mutation analysis. Exon 15, V600E which contains the BRAF mutation, was amplified by PCR with the foward primer AGGAAAGCATCTCACCTCATC and the reverse primerGATCACACCTGCCTTAAATTGC. The PCR parameters were as follows: 94˚C for 5 minutes, 35 cycles at 94˚C for 0.5 minutes, 60˚C for 0.5 minutes, and 72˚C for 10 minutes. The amplified products were purified with a QIA- GEN PCR purification kit and sequenced using the foward primer described previously with Big Dye Terminator (ABI Systems, Applied Biosystems, Foster City, CA), and an ABI PRISM 3100 Avant Genetic Analyzer (Perkin-Elmer). Data & statistical analysis For feature selection, LASSO logistic regression model was applied to the 730 texture features extracted from the US images, and a Radiomics Score was calculated for each patient using a linear combination of selected features weighted by the respective coefficients. Univariable and multivariable logistic regression analysis was performed to calculate the odds ratio with 95% confidence intervals (CI), including patient’s age, gender, tumor size, and radiomics score. For internal validation, bootstrap with 1,000 resampling was used. Calibra- tion curves were plotted to assess the calibration of the model built using the factors included, using the Hosmer Lemeshow test. Harrell’s C-index was measured to evaluate the model’s dis- crimination ability. R software (version 3.4.2, http://www.R-project.org) with the R package ‘glmnet’ was used for statistical analysis. Results Among the 527 patients diagnosed with PTC in this study, 428 (81.2%) were positive and 99 V600E (18.8%) were negative for BRAF mutation. Cancer subtype of the 527 masses were proven as follows: 493 (93.5%) as conventional PTC, 23 (4.4%) as follicular variant PTC, 7 (1.3%) as diffuse sclerosing variant PTC, 4 (0.8%) as oncocytic variant of PTC. Demographics according V600E to the presence of BRAF mutations are summarized in Table 1. Mean tumor size was sig- V600E nificantly smaller in nodules positive for BRAF mutation, 16.0±7.6 mm to 18.0±9.1 mm (P = 0.003). Similar tendency was observed in the training cohort (15.8±7.4 mm to 19.6±10.0 mm, P = 0.004), but not in the validation cohort or the conventional PTC <20-mm subgroup. Feature selection and calcuation of radiomics score Eight potential features were selected among 730 texture features in the training cohort with nonzero coefficients in the LASSO logistic regression model (Fig 2A and 2B). These 8 texture features were presented in the calculation formula below used to calculate the Radiomics PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 4 / 11 V600E Radiomics in predicting BRAF mutation V600E Table 1. Demographic features of the total thyroid cancers and conventional PTCs<20-mm according to the presence of BRAF mutation. Total (n = 527) P Training cohort (n = 387) P Validation cohort (n = 140) P V600E BRAF mutation Negative Positive Negative Positive Negative Positive (n = 99) (n = 428) (n = 68) (n = 319) (n = 31) (n = 109) Mean age (years) 38.4±13.2 42.7±13.8 0.284 38.3±13.4 43.1±13.9 0.009 38.8±13.1 41.7±13.4 0.288 <55 years 86 (86.9%) 334 (78.0%) 0.064 60 (87.0) 247 (77.7) 0.118 27 (87.1) 86 (78.9) 0.446 �55 years 13 (13.1%) 94 (22.0%) 9 (13.0) 71 (22.3) 4 (12.9) 23 (21.1) Gender 0.523 0.997 0.290 Men 79 (79.8%) 326 (76.2%) 15 (22.7) 72 (22.6) 5 (16.1) 30 (27.5) Women 20 (20.2%) 102 (23.8%) 54 (78.3) 246 (77.4) 26 (83.9) 79 (72.5) Mean size of tumor (mm) 18.0±9.1 16.0±7.6 0.003 19.6±10.0 15.8±7.4 0.004 15.2±7.8 16.2±7.6 0.511 <20mm 67 (67.7%) 341 (79.7%) 0.015 40 (58.0) 254 (79.9) <0.001 27 (87.1) 87 (79.8) 0.511 �20mm 32 (32.3%) 79 (20.3%) 29 (42.0) 64 (20.1) 4 (12.9) 22 (20.2) Radiomics Score (median, 1.486 (1.160, 1.704 (1.501, <0.001 1.615 (1.457, 1.672 (1.519, 0.238 interquartile range) 1.690) 1.836) 1.747) 1.832) Conventional PTC <20-mm (n = 389) Training cohort (n = 280) P Validation cohort (n = 109) P Negative Positive Negative Positive (n = 33) (n = 247) (n = 23) (n = 86) Mean age (years) 41.1±13.9 43.3±13.0 0.386 39.5±13.9 43.0±12.6 0.277 <55 years 27 (81.8) 194 (78.5) 0.837 19 (82.6) 68 (79.1) >0.999 �55 years 6 (18.2) 53 (21.5) 4 (17.4) 18 (20.9) Gender 0.151 0.587 Men 3 (9.1) 53 (21.5) 4 (17.4) 22 (25.6) Women 30 (90.9) 194 (78.5) 19 (82.6) 64 (74.4) Mean size of tumor (mm) 12.9±2.7 13.0±2.3 0.843 13.1±2.0 13.3±2.4 0.720 <20mm - - - - - - - �20mm - - - - - - - Radiomics Score (median, 1.887 (1.734, 2.117 (1.929, <0.001 2.089 (1.920, 2.104 (1.949, 0.827 interquartile range) 2.161) 2.257) 2.246) 2.223) https://doi.org/10.1371/journal.pone.0228968.t001 Score, Radiomics ScoreðtotalÞ ¼ 0:3715483 #0; 0:0179227 X mad 6 0 #0; 0:0202624 X sv 43 0 0:0000068 X HL ene 1 0 #0; 0:0000041 X HL rln 48 0 #0; 0:0769504 X LL uni 13 0 0:0013692 X LL lrlgle 54 0þ 0:0025444 X LL se 42 45þ 0:5554316 X LL se 42 90 For the conventional PTCs measuring <20-mm, 4 potential features were selected among the 730 texture features in the training cohort (Fig 2C and 2D). These 4 texture features were presented in the calculation formula below used to calculate the Radiomics Score (cPTC<20-mm), Radiomics ScoreðcPTC < 20#0; mmÞ ¼ #0; 2:2001791þ 11:4205518 X LH srlgle 52 0#0; 0:7666155 X LL uni 13 0 þ 0:8461400 X LL se 42 90#0; 0:0001180 X LL lrhgle 55 90 PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 5 / 11 V600E Radiomics in predicting BRAF mutation Fig 2. Texture feature selection using the least absolute shrinkage and selection operator (LASSO) logistic regression model. (A) Tuning parameter (lambda,λ) selection in the LASSO model used 10-fold cross validation for 527 thyroid cancers. The mean deviance (goodness-of-fit statistics, red dots) was plotted versus log(λ), error bars displaying the range of standard error. Dotted vertical lines were drawn at the point of minimum deviance (λ value = 0.03229), and at the point where maximumλ was obtained among errors smaller than the standard error of minimum deviance (λ value = 0.08984). (B) LASSO coefficient profiles of the 730 texture features. A coefficient profile was plotted versus log(λ). The gray vertical line was drawn at the value selected using 10-fold cross validation, where the optimalλ resulted in 8 nonzero coefficients. (C) Tuning parameter (lambda,λ) selection in the LASSO model used 10-fold cross validation for 389 conventional PTCs <20-mm. The mean deviance (goodness-of-fit statistics, red dots) was plotted versus log(λ), error bars displaying the range of standard error. Dotted vertical lines were drawn at the point of minimum deviance (λ value = 0.0329208), and at the point where maximumλ was obtained among errors smaller than the standard error of minimum deviance (λ value = 0.072595). (D) LASSO coefficient profiles plotted versus log(λ), gray vertical line was drawn at the value selected using 10-fold cross validation, where the optimalλ resulted in 4 nonzero coefficients. https://doi.org/10.1371/journal.pone.0228968.g002 Development, performance, and validation of prediction models Table 2 summarizes the results of univariable and multivariable logistic regression analysis for V600E predicting the presence of BRAF mutations. In the training cohort of the total thyroid cancers, tumor size and Radiomics Score were factors with statistical significance on univari- able analysis. Among the training cohort including conventional PTCs measuring <20-mm, Radiomics Score was the single factor showing statistical significance. In both total cancers and the conventional PTC<20-mm, Radiomics Score was the single factor showing significant V600E association to the presence of BRAF mutation on multivariable analysis (all P<0.05). PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 6 / 11 V600E Radiomics in predicting BRAF mutation V600E Table 2. Univariable and multivariable analysis in predicting the presence of BRAF mutation in the training cohort of the total thyroid cancers and conven- tional PTC<20-mm. Clinical features Total Univariable Multivariable OR 95% CI P OR 95% CI P Tumor size 0.953 0.924–0.981 0.001 1.018 0.977–1.060 0.394 Age (�55 years) 1.916 0.948–4.308 0.089 1.948 0.928–4.536 0.096 Gender 1.054 0.573–2.035 0.871 1.757 0.877–3.793 0.129 Radiomics score 6.099 3.124–12.723 <0.001 8.979 3.603–23.920 <0.001 Conventional PTCs<20-mm OR 95% CI P OR 95% CI P Tumor size 1.018 0.875–1.198 0.825 1.030 0.872–1.232 0.738 Age (�55 years) 1.229 0.512–3.431 0.665 1.282 0.495–3.943 0.632 Gender 2.732 0.926–11.701 0.108 4.281 1.166–27.327 0.060 Radiomics score 9.976 3.161–38.451 <0.001 11.279 3.624–44.121 <0.001 US: ultrasonography, PTC: papillary thyroid carcinoma, OR: Odds ratio, 95% CI: 95% confidence interval https://doi.org/10.1371/journal.pone.0228968.t002 V600E The calibration curve of the prediction model for the presence of BRAF mutation demonstrated good agreement between prediction and observation in the training cohort among the thyroid cancers. The Hosmer-Lemeshow test yielded statistics of P = 0.502, suggest- ing good calibration (Fig 3A). C-statistics for the training set was 0.718 (95% CI: 0.650–0.786), and 0.629 (95% CI: 0.516–0.742) for the validation set (Table 3). The calibration curve of V600E the prediction model for the presence of BRAF mutation among conventional PTCs <20-mm demonstrated good calibration, with the Hosmer-Lemeshow test yielding statistics of P = 0.257 (Fig 3B). C-statistics for training set among the conventional PTCs<20-mm was 0.729 (95% CI: 0.632–0.826), and 0.567 (95% CI: 0.434–0.699) for the validation set. Discussion One major challenge for thyroid cancer is how to distinguish patients who need aggressive treatment to survive to those who do not. There are no consistent predictors that reliably sorts out aggressive PTCs, and in addition to the lack of prospective data regarding appropriate treatment for PTCs due to its generally excellent survival [9], issues regarding overtreatment for low-risk patients who will not experience PTC-related mortality have surfaced and debated over the recent years. This reflects the need for a more effective and accurate biomarker in V600E predicting aggressive PTCs, including molecular analysis such as BRAF mutations. Muta- tion analysis requires invasive procedures such as biopsy or surgical resection to retrive speci- men to be analyzed. Among the non-invasive imaging biomarkers, radiomics is an emerging method that has the potential to predict molecular characteristics of tumors, using quantitative imaging features extracted using data-characterization algorithms. The most widely used imaging modality in radiomics has been computed tomography (CT) or magnetic resonance imaging (MRI), however, US is the most sensitive and accurate imaging modality for the thy- roid which we used in this study. For feature selection in obtaining a Radiomics Score, the LASSO logistic regression model was used, which enables selecting features based on their strength of association on univariable analysis, and combining the selected features into a radiomics signature [23]. The Radiomics Score obtained was the single factor showing significant association in pre- V600E dicting the presence of BRAF mutation in both univariable and multivariable analysis PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 7 / 11 V600E Radiomics in predicting BRAF mutation V600E Fig 3. Calibration plots of the grouped prediction models for the presence of BRAF mutation. For each plot, V600E the y-axis represents the actual probability of BRAF mutation, and the x-axis represents the predicted risk for V600E BRAF mutation. (A) Calibration plot for the total thyroid cancers and (B) conventional PTCs measuring <20-mm included in this study. https://doi.org/10.1371/journal.pone.0228968.g003 PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 8 / 11 V600E Radiomics in predicting BRAF mutation Table 3. Discrimination ability of the models in the total thyroid cancers and the conventional PTCs<20-mm. Total (n = 527) Conventional PTC<20-mm (n = 389) Original Internal validation Original Internal validation c-statistics 95% CI Boostrapped c-statistics 95% CI c-statistics 95% CI Boostrapped c-statistics 95% CI Training set 0.718 0.650–0.786 0.716 0.652–0.786 0.729 0.632–0.826 0.729 0.634–0.819 Validation set 0.629 0.516–0.742 0.566 0.430–0.701 https://doi.org/10.1371/journal.pone.0228968.t003 in the training cohorts (Table 3), showing good discrimination for thyroid cancers with V600E BRAF mutation in the training set (c-statistics 0.718 (95% CI: 0.650–0.786)), but with lower c-statistics for validation set (0.629 (95% CI: 0.516–0.742). Our results show that US- derived radiomics may have potential as a non-invasive biomarker, but currently does not V600E enable accurate prediction of the presence of BRAF mutation in PTCs. There have been other studies proving the potential of US-derived radiomics in predicting disease-free sur- vival or in predicting lymph node metastasis in patients diagnosed with PTC [24,25], but to the best of our knowledge, there are currently no studies using US radiomics features in pre- V600E dicting the presence of BRAF mutations in patients diagnosed with thyroid cancer. Fur- ther studies including larger number of cases are anticipated in the future to validate our results. V600E Among the clinical variables, higher rates of BRAF mutation was seen in thyroid cancers of smaller size, which showed significant association on univariable analysis. As the tumor size and subtype of thyroid cancer has been reported to have association to the V600E presence of BRAF mutation [21], we performed a subgroup analysis using a separate Radiomics Score calculated among the 730 texture features from a subset of 389 thyroid cancers confirmed as conventional PTC<20-mm. When using the Radiomics Score (cPTC<20-mm), similar results were obtained with the total 527 PTCs; Radiomics Score (cPTC<20-mm) was the single factor showing significance on both univariable/multivari- able analysis (all P<0.001), with c-statistics of 0.729 (95% CI: 0.632–0.826) for the training set, lower values for the validation set, 0.567 (95% CI: 0.434–0.699). This supports that US- V600E radiomics has limited value in predicting BRAF mutation in PTC patients, regardless of size. There are several limitations to this study. First, as the mutation analysis was performed in a selected group of PTC patients, results of our study does not represent mutation features of the general thyroid cancer population. Second, 81.2% of the PTCs in this study had V600E BRAF mutation analysis, which may have affected our results. PTCs among our popula- V600E tion has been known for its high prevalence for BRAF mutation [26], and results may have differed when conducted on different populations. Last, US images were used for fea- ture extraction in obtaining a Radiomics Score that may be used in prediction of the pres- V600E ence of BRAF mutation. Inherent observer variability of US compared to computed tomography (CT) or magnetic resonance imaging (MRI) may have affected our results, but since US is currently the generally applied imaging modality for detecting and differentiat- ing thyroid nodules, feature extraction from US images may be more appropriate in extract- ing radiomics data among thyroid imaging. Also, ROIs for feature extraction was obtained from one radiologist, and observer variability among different radiologists were not consid- ered in data analysis. In conclusion, our results show that radiomics features extracted from US has limited value V600E as a non-invasive biomarker for predicting the presence of BRAF mutation status of PTC regardless of size. PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 9 / 11 V600E Radiomics in predicting BRAF mutation Author Contributions Conceptualization: Jin Young Kwak. Data curation: Jung Hyun Yoon, Kyunghwa Han, Eunjung Lee, Jin Young Kwak. Formal analysis: Jung Hyun Yoon, Kyunghwa Han, Eunjung Lee. Funding acquisition: Jin Young Kwak. Investigation: Jung Hyun Yoon. Methodology: Jin Young Kwak. Project administration: Jin Young Kwak. Resources: Jandee Lee, Kee Hyun Nam, Jin Young Kwak. Software: Eunjung Lee. Supervision: Eun-Kyung Kim, Hee Jung Moon, Vivian Youngjean Park, Jin Young Kwak. Writing – original draft: Jung Hyun Yoon, Jin Young Kwak. Writing – review & editing: Jung Hyun Yoon, Kyunghwa Han, Eunjung Lee, Jandee Lee, Eun-Kyung Kim, Hee Jung Moon, Vivian Youngjean Park, Kee Hyun Nam, Jin Young Kwak. References 1. Chen AY, Jemal A, Ward EM. Increasing incidence of differentiated thyroid cancer in the United States, 1988–2005. Cancer. 2009; 115: 3801–3807. https://doi.org/10.1002/cncr.24416 PMID: 19598221 2. George JR, Henderson YC, Williams MD, Roberts DB, Hei H, Lai SY, et al. Association of TERT Pro- moter Mutation, But Not BRAF Mutation, With Increased Mortality in PTC. 3. Magreni A, Bann DV, Schubart JR, Goldenberg D. The effects of race and ethnicity on thyroid cancer incidence. JAMA Otolaryngol Head Neck Surg. 2015; 141: 319–323. https://doi.org/10.1001/jamaoto. 2014.3740 PMID: 25654447 4. Davies L, Welch HG. Increasing incidence of thyroid cancer in the United States, 1973–2002. JAMA. 2006; 295: 2164–2167. https://doi.org/10.1001/jama.295.18.2164 PMID: 16684987 5. Enewold L, Zhu K, Ron E, Marrogi AJ, Stojadinovic A, Peoples GE, et al. Rising thyroid cancer inci- dence in the United States by demographic and tumor characteristics, 1980–2005. Cancer Epidemiol Biomarkers Prev. 2009; 18: 784–791. https://doi.org/10.1158/1055-9965.EPI-08-0960 PMID: 6. Yin DT, Yu K, Lu RQ, Li X, Xu J, Lei M, et al. Clinicopathological significance of TERT promoter muta- tion in papillary thyroid carcinomas: a systematic review and meta-analysis. Clin Endocrinol (Oxf). 2016; 85: 299–305. https://doi.org/10.1111/cen.13017 PMID: 26732020 7. Noone AM, Cronin KA, Altekruse SF, Howlader N, Lewis DR, Petkov VI, et al. Cancer Incidence and Survival Trends by Subtype Using Data from the Surveillance Epidemiology and End Results Program, 1992–2013. Cancer Epidemiol Biomarkers Prev. 2017; 26: 632–641. https://doi.org/10.1158/1055- 9965.EPI-16-0520 PMID: 27956436 8. Xing M, Alzahrani AS, Carson KA, Viola D, Elisei R, Bendlova B, et al. Association between BRAF V600E mutation and mortality in patients with papillary thyroid cancer. JAMA. 2013; 309: 1493–1501. https://doi.org/10.1001/jama.2013.3190 PMID: 23571588 9. Brown RL, de Souza JA, Cohen EE. Thyroid cancer: burden of illness and management of disease. J Cancer. 2011; 2: 193–199 https://doi.org/10.7150/jca.2.193 PMID: 21509149 10. Mazzaferri EL, Jhiang SM. Long-term impact of initial surgical and medical therapy on papillary and fol- licular thyroid cancer. Am J Med. 1994; 97: 418–428 https://doi.org/10.1016/0002-9343(94)90321-2 PMID: 7977430 11. Xing MZ, Liu RY, Liu XL, Murugan AK, Zhu GW, Zeiger MA, et al. BRAF V600E and TERT Promoter Mutations Cooperatively Identify the Most Aggressive Papillary Thyroid Cancer With Highest Recur- rence. Journal of Clinical Oncology. 2014; 32: 2718–+. https://doi.org/10.1200/JCO.2014.55.5094 PMID: 25024077 PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 10 / 11 V600E Radiomics in predicting BRAF mutation 12. Loh KC, Greenspan FS, Gee L, Miller TR, Yeo PP. Pathological tumor-node-metastasis (pTNM) staging for papillary and follicular thyroid carcinomas: a retrospective analysis of 700 patients. J Clin Endocrinol Metab. 1997; 82: 3553–3562. https://doi.org/10.1210/jcem.82.11.4373 PMID: 9360506 13. Penna GC, Vaisman F, Vaisman M, Sobrinho-Simoes M, Soares P. Molecular Markers Involved in Tumorigenesis of Thyroid Carcinoma: Focus on Aggressive Histotypes. Cytogenet Genome Res. 2016; 150: 194–207. https://doi.org/10.1159/000456576 PMID: 28231576 14. Melo M, Gaspar da Rocha A, Batista R, Vinagre J, Martins MJ, Costa G, et al. TERT, BRAF, and NRAS in Primary Thyroid Cancer and Metastatic Disease. 15. Kim TH, Park YJ, Lim JA, Ahn HY, Lee EK, Lee YJ, et al. The association of the BRAF(V600E) mutation with prognostic factors and poor clinical outcome in papillary thyroid cancer: A Meta-Analysis. Cancer. 2011. https://doi.org/10.1002/cncr.26500 PMID: 21882184 16. Liu R, Bishop J, Zhu G, Zhang T, Ladenson PW, Xing M. Mortality Risk Stratification by Combining BRAF V600E and TERT Promoter Mutations in Papillary Thyroid Cancer: Genetic Duet of BRAF and TERT Promoter Mutations in Thyroid Cancer Mortality. LID - https://doi.org/10.1001/jamaoncol.2016. 3288 PMID: 27581851 17. Gevaert O, Echegaray S, Khuong A, Hoang CD, Shrager JB, Jensen KC, et al. Predictive radioge- nomics modeling of EGFR mutation status in lung cancer. 18. Braman NM, Etesami M, Prasanna P, Dubchuk C, Gilmore H, Tiwari P, et al. Intratumoral and peritu- moral radiomics for the pretreatment prediction of pathological complete response to neoadjuvant che- motherapy based on breast DCE-MRI. 19. Zhang B, Tian J, Dong D, Gu D, Dong Y, Zhang L, et al. Radiomics Features of Multiparametric MRI as Novel Prognostic Factors in Advanced Nasopharyngeal Carcinoma. LID - https://doi.org/10.1158/1078- 0432.CCR-16-2910 PMID: 28280088 20. Yang L, Dong D, Fang M, Zhu Y, Zang Y, Liu Z, et al. Can CT-based radiomics signature predict KRAS/ NRAS/BRAF mutations in colorectal cancer? Eur Radiol. 2018; 28: 2058–2067. https://doi.org/10.1007/ s00330-017-5146-8 PMID: 29335867 21. Li C, Lee KC, Schneider EB, Zeiger MA. BRAF V600E mutation and its association with clinicopatholog- ical features of papillary thyroid cancer: a meta-analysis. J Clin Endocrinol Metab. 2012; 97: 4559– 4570. https://doi.org/10.1210/jc.2012-2104 PMID: 23055546 22. Aerts HJ, Velazquez ER, Leijenaar RT, Parmar C, Grossmann P, Carvalho S, et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun. 2014; 5: 4006. https://doi.org/10.1038/ncomms5006 PMID: 24892406 23. Harell F. Regression Modeling Strategies With Applicators to Linear Models, Logistic regression, and Survival Analysis. New York, NY: Springer; 2015. 24. Park VY, Han K, Lee E, Kim EK, Moon HJ, Yoon JH, et al. Association Between Radiomics Signature and Disease-Free Survival in Conventional Papillary Thyroid Carcinoma. Sci Rep. 2019; 9: 4501. https://doi.org/10.1038/s41598-018-37748-4 PMID: 30872763 25. Liu T, Zhou S, Yu J, Guo Y, Wang Y, Zhou J, et al. Prediction of Lymph Node Metastasis in Patients With Papillary Thyroid Carcinoma: A Radiomics Method Based on Preoperative Ultrasound Images. Technol Cancer Res Treat. 2019; 18: 1533033819831713. https://doi.org/10.1177/1533033819831713 PMID: 30890092 26. Kim KH, Kang DW, Kim SH, Seong IO, Kang DY. Mutations of the BRAF gene in papillary thyroid carci- noma in a Korean population. Yonsei Med J. 2004; 45: 818–821. https://doi.org/10.3349/ymj.2004.45.5. 818 PMID: 15515191 PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 11 / 11 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png PLoS ONE Public Library of Science (PLoS) Journal

Radiomics in predicting mutation status for thyroid cancer: A preliminary study using radiomics features for predicting BRAFV600E mutations in papillary thyroid carcinoma

PLoS ONE, Volume 15 (2) – Feb 13, 2020

Loading next page...
 
/lp/public-library-of-science-plos-journal/radiomics-in-predicting-mutation-status-for-thyroid-cancer-a-ZaEyBMwClB
Publisher
Public Library of Science (PLoS) Journal
Copyright
Copyright: © 2020 Yoon et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: Data cannot be shared publicly because of patient confidentiality issues. The institutional review board of Severance Hospital, Yonsei University has given approval for this study, recommending patient confidentiality when using image data. contact: hpc@yuhs.ac. Funding: This work was supported by the National Research Foundation of Korea(NRF) grant funded by the Korea government(MSIT) (2019R1A2C1002375). This study was also supported by a CMB-Yuhan research grant of Yonsei University College of Medicine (6-2017-0170). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist.
eISSN
1932-6203
DOI
10.1371/journal.pone.0228968
Publisher site
See Article on Publisher Site

Abstract

OPENACCESS Citation: Yoon JH, Han K, Lee E, Lee J, Kim E-K, Moon HJ, et al. (2020) Radiomics in predicting Purpose mutation status for thyroid cancer: A preliminary To evaluate whether if ultrasonography (US)-based radiomics enables prediction of the study using radiomics features for predicting V600E V600E presence of BRAF mutations among patients diagnosed as papillary thyroid carcni- BRAF mutations in papillary thyroid carcinoma. PLoS ONE 15(2): e0228968. https:// noma (PTC). doi.org/10.1371/journal.pone.0228968 Editor: Jeeun Kang, Johns Hopkins University, Methods UNITED STATES From December 2015 to May 2017, 527 patients who had been treated surgically for PTC V600E Received: August 2, 2019 were included (training: 387, validation: 140). All patients had BRAF mutation analysis Accepted: January 27, 2020 performed on surgical specimen. Feature extraction was performed using preoperative US images of the 527 patients (mean size of PTC: 16.4mm±7.9, range, 10–85 mm). A Radio- Published: February 13, 2020 mics Score was generated by using the least absolute shrinkage and selection operator Copyright:© 2020 Yoon et al. This is an open (LASSO) regression model. Univariable/multivariable logistic regression analysis was per- access article distributed under the terms of the V600E formed to evaluate the factors including Radiomics Score in predicting BRAF mutation. Creative Commons Attribution License, which permits unrestricted use, distribution, and Subgroup analysis including conventional PTC<20-mm (n = 389) was performed (training: reproduction in any medium, provided the original 280, validation: 109). author and source are credited. Data Availability Statement: Data cannot be Results shared publicly because of patient confidentiality Of the 527 patients diagnosed with PTC, 428 (81.2%) were positive and 99 (18.8%) were issues. The institutional review board of Severance V600E Hospital, Yonsei University has given approval for negative for BRAF mutation. In both total 527 cancers and 389 conventional PTC<20- this study, recommending patient confidentiality mm, Radiomics Score was the single factor showing significant association to the presence when using image data. contact: hpc@yuhs.ac. V600E of BRAF mutation on multivariable analysis (all P<0.05). C-statistics for the validation Funding: This work was supported by the National set in the total cancers and the conventional PTCs<20-mm were lower than that of the train- Research Foundation of Korea(NRF) grant funded ing set: 0.629 (95% CI: 0.516–0.742) to 0.718 (95% CI: 0.650–0.786), and 0.567 (95% CI: by the Korea government(MSIT) (2019R1A2C1002375). This study was also 0.434–0.699) to 0.729 (95% CI: 0.632–0.826), respectively. PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 1 / 11 V600E Radiomics in predicting BRAF mutation supported by a CMB-Yuhan research grant of Conclusion Yonsei University College of Medicine (6-2017- Radiomics features extracted from US has limited value as a non-invasive biomarker for 0170). The funders had no role in study design, V600E data collection and analysis, decision to publish, or predicting the presence of BRAF mutation status of PTC regardless of size. preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. Introduction During the past decade, the incidence of thyroid cancer has rapidly increased worldwide, regardless of the demographic groups [1–3]. The majority of thyroid cancers that are being newly detected are papillary thyroid cancers (PTC) [4,5], a subtype consisting of more than 80% of all differentiated thyroid carcinomas [5,6]. In general, PTC is known to have excellent patient outcomes, 5-year survival rates approaching 98–99% [7,8], but approximately 10–15% of patients had aggressive tumor behavior, local recurrence/distant metastasis after treatment or mortality [9–12]. At present, even with the well-known poor prognostic factors such as age over 45 years, male gender, radioactive iodine resistance [9], it is difficult to descriminate to predict which patient has more aggressive forms of PTCs, and effort has been made using vari- ous biomarkers in predicting PTC patients with poor outcome. With the advancement in molecular genetics, various genetic alterations have been revealed and used as an adjunctive diagnostic method or for predicting patient prognosis [8,13,14]. V600E BRAF mutation, the most frequent oncogene in PTC, has been reported to be associated with aggressive clinical features such as large tumor size, extrathyroidal extension and presence of lymph node metastasis [8,14–16], leading to recurrence or mortality. But even with the abil- ity of either mutation in detecting aggressive cancer types, genetic analysis requires specimen tissue for analysis, mostly obtained from invasive surgical procedures. Aside from the informa- tion obtained from conventional imaging, radiomics, using data extracted from medical images converted into high-dimensional, mineable, and quantitative imaging features has been applied to revealing tumor physiology. Other studies have linked imaging features to molecular properties of tumors among various organs [17–20], but to the best of our knowl- edge, no studies have applied radiomics in predicting molecular status of thyroid cancer that can be used in predicting tumor aggressiveness. Based on this, we evaluated whether if ultraso- V600E nography (US)-based radiomics enables prediction of the presence of BRAF mutations among patients diagnosed as PTC. Materials & methods This retrospective study has been approved by the institutional review board (IRB) of Sever- ance Hospital, Yonsei University (approval number: 4-2018-0172), with a waiver for patient consent due to the retrospective study design. Signed informed consent was obtained from all patients prior to biopsy or surgical procedures. Images used for data extraction were fully anonymized before data processing according to the instructions of our IRB. Patients We included 527 patients who had been treated surgically with cytologically-proven or suspi- cious thyroid cancer between December 2015 to May 2017 at Severance Hospital, Seoul, V600E Korea. All patients had BRAF mutation analysis performed on surgical specimen. The 387 consecutive patients who had surgery from December 2015 to December 2016 were used as the training cohort: 300 women, 87 men, mean age, 42.1 years±14.0 (range, 15–82 years). PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 2 / 11 V600E Radiomics in predicting BRAF mutation The 140 consecutive patients who had surgery from January 2017 to May 2017 were used as the validation cohort: 105 women, 35 men, mean age, 41.3 years±13.4 (range, 15–74 years). Mean age of the 527 patients were 41.9 years±13.8 (range, 15–82 years). Mean size of the thy- roid masses was 16.4 mm±7.9 (range, 10–85 mm). V600E As the frequency of BRAF mutation has been reported to be associated with tumor size and conventional PTC [21], subgroup analysis was performed including thyroid cancers confirmed as conventional PTCs measuring <20-mm (n = 389). Mean age of the 389 patients was 42.9 years±13.1 (range, 15–80 years). Mean size of the conventional PTCs was 14.9 mm ±4.6 (range, 10–19 mm). Clinicopathologic data regarding tumor size, lymph node metastasis were obtained from review of medical records. Imaging features of the thyroid masses used for analysis were obtained from an institutional database. US image selection and feature extraction One radiologist (J.Y.K.) reviewed the preoperative US examinations of the 527 patients on the picture archiving and communication system (PACS) and selected representative transverse or longitudinal images of the tumor. The selected representative images were converted into JPEG files for manual segmentation. One radiologist (J.H.Y.) who had 9 years of experience in thyroid imaging manually set a region-of-interest (ROI) along the boundary of the selected tumor using Paint software of Windows (Fig 1). Since ROI marking with colored brush using Paint software alters original intensities in image, the manual ROI segmentation is conducted over the duplicate images of collected JPEG files. Before starting ROI extracting procedure, all images were normalized for fair comparison. First the location information of ROI marking (coordinate information of red curves in Fig 1) was sought and then applied to the original JPEG image to extract ROI only. This procedure ensures that the original intensity of the Fig 1. Representative image of tumor segmentation using thyroid US. A diagonal region-of-interest (ROI) was drawn along the tumor border (red line) for feature extraction. https://doi.org/10.1371/journal.pone.0228968.g001 PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 3 / 11 V600E Radiomics in predicting BRAF mutation image was not affected by the ROI extraction process. Once ROI was extracted, a total of 730 feature information were gathered. The 730 features include the first order statistics (energy, entropy, kurtosis, skewness and so on), the second order statistics (the gray level co-occurrence matrix (GLCM) and gray level run-length matrix (GLRLM) were established and the corre- sponding features were extracted), and features from four discrete one-level wavelet decompo- sitions. The detailed calculation for these features can be found in [22]. To obtain the feature quantities, the house code in MATLAB 2018b was used. Here, 256 bins using a bin with of 1 were utilized for intensity histogram and 4 angles of 0, 45, 90, and 135 degrees were utilized for GLCM and GLRLM anaylsis. V600E BRAF mutational analysis Direct DNA sequencing was used for the surgical specimen in mutation analysis. Exon 15, V600E which contains the BRAF mutation, was amplified by PCR with the foward primer AGGAAAGCATCTCACCTCATC and the reverse primerGATCACACCTGCCTTAAATTGC. The PCR parameters were as follows: 94˚C for 5 minutes, 35 cycles at 94˚C for 0.5 minutes, 60˚C for 0.5 minutes, and 72˚C for 10 minutes. The amplified products were purified with a QIA- GEN PCR purification kit and sequenced using the foward primer described previously with Big Dye Terminator (ABI Systems, Applied Biosystems, Foster City, CA), and an ABI PRISM 3100 Avant Genetic Analyzer (Perkin-Elmer). Data & statistical analysis For feature selection, LASSO logistic regression model was applied to the 730 texture features extracted from the US images, and a Radiomics Score was calculated for each patient using a linear combination of selected features weighted by the respective coefficients. Univariable and multivariable logistic regression analysis was performed to calculate the odds ratio with 95% confidence intervals (CI), including patient’s age, gender, tumor size, and radiomics score. For internal validation, bootstrap with 1,000 resampling was used. Calibra- tion curves were plotted to assess the calibration of the model built using the factors included, using the Hosmer Lemeshow test. Harrell’s C-index was measured to evaluate the model’s dis- crimination ability. R software (version 3.4.2, http://www.R-project.org) with the R package ‘glmnet’ was used for statistical analysis. Results Among the 527 patients diagnosed with PTC in this study, 428 (81.2%) were positive and 99 V600E (18.8%) were negative for BRAF mutation. Cancer subtype of the 527 masses were proven as follows: 493 (93.5%) as conventional PTC, 23 (4.4%) as follicular variant PTC, 7 (1.3%) as diffuse sclerosing variant PTC, 4 (0.8%) as oncocytic variant of PTC. Demographics according V600E to the presence of BRAF mutations are summarized in Table 1. Mean tumor size was sig- V600E nificantly smaller in nodules positive for BRAF mutation, 16.0±7.6 mm to 18.0±9.1 mm (P = 0.003). Similar tendency was observed in the training cohort (15.8±7.4 mm to 19.6±10.0 mm, P = 0.004), but not in the validation cohort or the conventional PTC <20-mm subgroup. Feature selection and calcuation of radiomics score Eight potential features were selected among 730 texture features in the training cohort with nonzero coefficients in the LASSO logistic regression model (Fig 2A and 2B). These 8 texture features were presented in the calculation formula below used to calculate the Radiomics PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 4 / 11 V600E Radiomics in predicting BRAF mutation V600E Table 1. Demographic features of the total thyroid cancers and conventional PTCs<20-mm according to the presence of BRAF mutation. Total (n = 527) P Training cohort (n = 387) P Validation cohort (n = 140) P V600E BRAF mutation Negative Positive Negative Positive Negative Positive (n = 99) (n = 428) (n = 68) (n = 319) (n = 31) (n = 109) Mean age (years) 38.4±13.2 42.7±13.8 0.284 38.3±13.4 43.1±13.9 0.009 38.8±13.1 41.7±13.4 0.288 <55 years 86 (86.9%) 334 (78.0%) 0.064 60 (87.0) 247 (77.7) 0.118 27 (87.1) 86 (78.9) 0.446 �55 years 13 (13.1%) 94 (22.0%) 9 (13.0) 71 (22.3) 4 (12.9) 23 (21.1) Gender 0.523 0.997 0.290 Men 79 (79.8%) 326 (76.2%) 15 (22.7) 72 (22.6) 5 (16.1) 30 (27.5) Women 20 (20.2%) 102 (23.8%) 54 (78.3) 246 (77.4) 26 (83.9) 79 (72.5) Mean size of tumor (mm) 18.0±9.1 16.0±7.6 0.003 19.6±10.0 15.8±7.4 0.004 15.2±7.8 16.2±7.6 0.511 <20mm 67 (67.7%) 341 (79.7%) 0.015 40 (58.0) 254 (79.9) <0.001 27 (87.1) 87 (79.8) 0.511 �20mm 32 (32.3%) 79 (20.3%) 29 (42.0) 64 (20.1) 4 (12.9) 22 (20.2) Radiomics Score (median, 1.486 (1.160, 1.704 (1.501, <0.001 1.615 (1.457, 1.672 (1.519, 0.238 interquartile range) 1.690) 1.836) 1.747) 1.832) Conventional PTC <20-mm (n = 389) Training cohort (n = 280) P Validation cohort (n = 109) P Negative Positive Negative Positive (n = 33) (n = 247) (n = 23) (n = 86) Mean age (years) 41.1±13.9 43.3±13.0 0.386 39.5±13.9 43.0±12.6 0.277 <55 years 27 (81.8) 194 (78.5) 0.837 19 (82.6) 68 (79.1) >0.999 �55 years 6 (18.2) 53 (21.5) 4 (17.4) 18 (20.9) Gender 0.151 0.587 Men 3 (9.1) 53 (21.5) 4 (17.4) 22 (25.6) Women 30 (90.9) 194 (78.5) 19 (82.6) 64 (74.4) Mean size of tumor (mm) 12.9±2.7 13.0±2.3 0.843 13.1±2.0 13.3±2.4 0.720 <20mm - - - - - - - �20mm - - - - - - - Radiomics Score (median, 1.887 (1.734, 2.117 (1.929, <0.001 2.089 (1.920, 2.104 (1.949, 0.827 interquartile range) 2.161) 2.257) 2.246) 2.223) https://doi.org/10.1371/journal.pone.0228968.t001 Score, Radiomics ScoreðtotalÞ ¼ 0:3715483 #0; 0:0179227 X mad 6 0 #0; 0:0202624 X sv 43 0 0:0000068 X HL ene 1 0 #0; 0:0000041 X HL rln 48 0 #0; 0:0769504 X LL uni 13 0 0:0013692 X LL lrlgle 54 0þ 0:0025444 X LL se 42 45þ 0:5554316 X LL se 42 90 For the conventional PTCs measuring <20-mm, 4 potential features were selected among the 730 texture features in the training cohort (Fig 2C and 2D). These 4 texture features were presented in the calculation formula below used to calculate the Radiomics Score (cPTC<20-mm), Radiomics ScoreðcPTC < 20#0; mmÞ ¼ #0; 2:2001791þ 11:4205518 X LH srlgle 52 0#0; 0:7666155 X LL uni 13 0 þ 0:8461400 X LL se 42 90#0; 0:0001180 X LL lrhgle 55 90 PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 5 / 11 V600E Radiomics in predicting BRAF mutation Fig 2. Texture feature selection using the least absolute shrinkage and selection operator (LASSO) logistic regression model. (A) Tuning parameter (lambda,λ) selection in the LASSO model used 10-fold cross validation for 527 thyroid cancers. The mean deviance (goodness-of-fit statistics, red dots) was plotted versus log(λ), error bars displaying the range of standard error. Dotted vertical lines were drawn at the point of minimum deviance (λ value = 0.03229), and at the point where maximumλ was obtained among errors smaller than the standard error of minimum deviance (λ value = 0.08984). (B) LASSO coefficient profiles of the 730 texture features. A coefficient profile was plotted versus log(λ). The gray vertical line was drawn at the value selected using 10-fold cross validation, where the optimalλ resulted in 8 nonzero coefficients. (C) Tuning parameter (lambda,λ) selection in the LASSO model used 10-fold cross validation for 389 conventional PTCs <20-mm. The mean deviance (goodness-of-fit statistics, red dots) was plotted versus log(λ), error bars displaying the range of standard error. Dotted vertical lines were drawn at the point of minimum deviance (λ value = 0.0329208), and at the point where maximumλ was obtained among errors smaller than the standard error of minimum deviance (λ value = 0.072595). (D) LASSO coefficient profiles plotted versus log(λ), gray vertical line was drawn at the value selected using 10-fold cross validation, where the optimalλ resulted in 4 nonzero coefficients. https://doi.org/10.1371/journal.pone.0228968.g002 Development, performance, and validation of prediction models Table 2 summarizes the results of univariable and multivariable logistic regression analysis for V600E predicting the presence of BRAF mutations. In the training cohort of the total thyroid cancers, tumor size and Radiomics Score were factors with statistical significance on univari- able analysis. Among the training cohort including conventional PTCs measuring <20-mm, Radiomics Score was the single factor showing statistical significance. In both total cancers and the conventional PTC<20-mm, Radiomics Score was the single factor showing significant V600E association to the presence of BRAF mutation on multivariable analysis (all P<0.05). PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 6 / 11 V600E Radiomics in predicting BRAF mutation V600E Table 2. Univariable and multivariable analysis in predicting the presence of BRAF mutation in the training cohort of the total thyroid cancers and conven- tional PTC<20-mm. Clinical features Total Univariable Multivariable OR 95% CI P OR 95% CI P Tumor size 0.953 0.924–0.981 0.001 1.018 0.977–1.060 0.394 Age (�55 years) 1.916 0.948–4.308 0.089 1.948 0.928–4.536 0.096 Gender 1.054 0.573–2.035 0.871 1.757 0.877–3.793 0.129 Radiomics score 6.099 3.124–12.723 <0.001 8.979 3.603–23.920 <0.001 Conventional PTCs<20-mm OR 95% CI P OR 95% CI P Tumor size 1.018 0.875–1.198 0.825 1.030 0.872–1.232 0.738 Age (�55 years) 1.229 0.512–3.431 0.665 1.282 0.495–3.943 0.632 Gender 2.732 0.926–11.701 0.108 4.281 1.166–27.327 0.060 Radiomics score 9.976 3.161–38.451 <0.001 11.279 3.624–44.121 <0.001 US: ultrasonography, PTC: papillary thyroid carcinoma, OR: Odds ratio, 95% CI: 95% confidence interval https://doi.org/10.1371/journal.pone.0228968.t002 V600E The calibration curve of the prediction model for the presence of BRAF mutation demonstrated good agreement between prediction and observation in the training cohort among the thyroid cancers. The Hosmer-Lemeshow test yielded statistics of P = 0.502, suggest- ing good calibration (Fig 3A). C-statistics for the training set was 0.718 (95% CI: 0.650–0.786), and 0.629 (95% CI: 0.516–0.742) for the validation set (Table 3). The calibration curve of V600E the prediction model for the presence of BRAF mutation among conventional PTCs <20-mm demonstrated good calibration, with the Hosmer-Lemeshow test yielding statistics of P = 0.257 (Fig 3B). C-statistics for training set among the conventional PTCs<20-mm was 0.729 (95% CI: 0.632–0.826), and 0.567 (95% CI: 0.434–0.699) for the validation set. Discussion One major challenge for thyroid cancer is how to distinguish patients who need aggressive treatment to survive to those who do not. There are no consistent predictors that reliably sorts out aggressive PTCs, and in addition to the lack of prospective data regarding appropriate treatment for PTCs due to its generally excellent survival [9], issues regarding overtreatment for low-risk patients who will not experience PTC-related mortality have surfaced and debated over the recent years. This reflects the need for a more effective and accurate biomarker in V600E predicting aggressive PTCs, including molecular analysis such as BRAF mutations. Muta- tion analysis requires invasive procedures such as biopsy or surgical resection to retrive speci- men to be analyzed. Among the non-invasive imaging biomarkers, radiomics is an emerging method that has the potential to predict molecular characteristics of tumors, using quantitative imaging features extracted using data-characterization algorithms. The most widely used imaging modality in radiomics has been computed tomography (CT) or magnetic resonance imaging (MRI), however, US is the most sensitive and accurate imaging modality for the thy- roid which we used in this study. For feature selection in obtaining a Radiomics Score, the LASSO logistic regression model was used, which enables selecting features based on their strength of association on univariable analysis, and combining the selected features into a radiomics signature [23]. The Radiomics Score obtained was the single factor showing significant association in pre- V600E dicting the presence of BRAF mutation in both univariable and multivariable analysis PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 7 / 11 V600E Radiomics in predicting BRAF mutation V600E Fig 3. Calibration plots of the grouped prediction models for the presence of BRAF mutation. For each plot, V600E the y-axis represents the actual probability of BRAF mutation, and the x-axis represents the predicted risk for V600E BRAF mutation. (A) Calibration plot for the total thyroid cancers and (B) conventional PTCs measuring <20-mm included in this study. https://doi.org/10.1371/journal.pone.0228968.g003 PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 8 / 11 V600E Radiomics in predicting BRAF mutation Table 3. Discrimination ability of the models in the total thyroid cancers and the conventional PTCs<20-mm. Total (n = 527) Conventional PTC<20-mm (n = 389) Original Internal validation Original Internal validation c-statistics 95% CI Boostrapped c-statistics 95% CI c-statistics 95% CI Boostrapped c-statistics 95% CI Training set 0.718 0.650–0.786 0.716 0.652–0.786 0.729 0.632–0.826 0.729 0.634–0.819 Validation set 0.629 0.516–0.742 0.566 0.430–0.701 https://doi.org/10.1371/journal.pone.0228968.t003 in the training cohorts (Table 3), showing good discrimination for thyroid cancers with V600E BRAF mutation in the training set (c-statistics 0.718 (95% CI: 0.650–0.786)), but with lower c-statistics for validation set (0.629 (95% CI: 0.516–0.742). Our results show that US- derived radiomics may have potential as a non-invasive biomarker, but currently does not V600E enable accurate prediction of the presence of BRAF mutation in PTCs. There have been other studies proving the potential of US-derived radiomics in predicting disease-free sur- vival or in predicting lymph node metastasis in patients diagnosed with PTC [24,25], but to the best of our knowledge, there are currently no studies using US radiomics features in pre- V600E dicting the presence of BRAF mutations in patients diagnosed with thyroid cancer. Fur- ther studies including larger number of cases are anticipated in the future to validate our results. V600E Among the clinical variables, higher rates of BRAF mutation was seen in thyroid cancers of smaller size, which showed significant association on univariable analysis. As the tumor size and subtype of thyroid cancer has been reported to have association to the V600E presence of BRAF mutation [21], we performed a subgroup analysis using a separate Radiomics Score calculated among the 730 texture features from a subset of 389 thyroid cancers confirmed as conventional PTC<20-mm. When using the Radiomics Score (cPTC<20-mm), similar results were obtained with the total 527 PTCs; Radiomics Score (cPTC<20-mm) was the single factor showing significance on both univariable/multivari- able analysis (all P<0.001), with c-statistics of 0.729 (95% CI: 0.632–0.826) for the training set, lower values for the validation set, 0.567 (95% CI: 0.434–0.699). This supports that US- V600E radiomics has limited value in predicting BRAF mutation in PTC patients, regardless of size. There are several limitations to this study. First, as the mutation analysis was performed in a selected group of PTC patients, results of our study does not represent mutation features of the general thyroid cancer population. Second, 81.2% of the PTCs in this study had V600E BRAF mutation analysis, which may have affected our results. PTCs among our popula- V600E tion has been known for its high prevalence for BRAF mutation [26], and results may have differed when conducted on different populations. Last, US images were used for fea- ture extraction in obtaining a Radiomics Score that may be used in prediction of the pres- V600E ence of BRAF mutation. Inherent observer variability of US compared to computed tomography (CT) or magnetic resonance imaging (MRI) may have affected our results, but since US is currently the generally applied imaging modality for detecting and differentiat- ing thyroid nodules, feature extraction from US images may be more appropriate in extract- ing radiomics data among thyroid imaging. Also, ROIs for feature extraction was obtained from one radiologist, and observer variability among different radiologists were not consid- ered in data analysis. In conclusion, our results show that radiomics features extracted from US has limited value V600E as a non-invasive biomarker for predicting the presence of BRAF mutation status of PTC regardless of size. PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 9 / 11 V600E Radiomics in predicting BRAF mutation Author Contributions Conceptualization: Jin Young Kwak. Data curation: Jung Hyun Yoon, Kyunghwa Han, Eunjung Lee, Jin Young Kwak. Formal analysis: Jung Hyun Yoon, Kyunghwa Han, Eunjung Lee. Funding acquisition: Jin Young Kwak. Investigation: Jung Hyun Yoon. Methodology: Jin Young Kwak. Project administration: Jin Young Kwak. Resources: Jandee Lee, Kee Hyun Nam, Jin Young Kwak. Software: Eunjung Lee. Supervision: Eun-Kyung Kim, Hee Jung Moon, Vivian Youngjean Park, Jin Young Kwak. Writing – original draft: Jung Hyun Yoon, Jin Young Kwak. Writing – review & editing: Jung Hyun Yoon, Kyunghwa Han, Eunjung Lee, Jandee Lee, Eun-Kyung Kim, Hee Jung Moon, Vivian Youngjean Park, Kee Hyun Nam, Jin Young Kwak. References 1. Chen AY, Jemal A, Ward EM. Increasing incidence of differentiated thyroid cancer in the United States, 1988–2005. Cancer. 2009; 115: 3801–3807. https://doi.org/10.1002/cncr.24416 PMID: 19598221 2. George JR, Henderson YC, Williams MD, Roberts DB, Hei H, Lai SY, et al. Association of TERT Pro- moter Mutation, But Not BRAF Mutation, With Increased Mortality in PTC. 3. Magreni A, Bann DV, Schubart JR, Goldenberg D. The effects of race and ethnicity on thyroid cancer incidence. JAMA Otolaryngol Head Neck Surg. 2015; 141: 319–323. https://doi.org/10.1001/jamaoto. 2014.3740 PMID: 25654447 4. Davies L, Welch HG. Increasing incidence of thyroid cancer in the United States, 1973–2002. JAMA. 2006; 295: 2164–2167. https://doi.org/10.1001/jama.295.18.2164 PMID: 16684987 5. Enewold L, Zhu K, Ron E, Marrogi AJ, Stojadinovic A, Peoples GE, et al. Rising thyroid cancer inci- dence in the United States by demographic and tumor characteristics, 1980–2005. Cancer Epidemiol Biomarkers Prev. 2009; 18: 784–791. https://doi.org/10.1158/1055-9965.EPI-08-0960 PMID: 6. Yin DT, Yu K, Lu RQ, Li X, Xu J, Lei M, et al. Clinicopathological significance of TERT promoter muta- tion in papillary thyroid carcinomas: a systematic review and meta-analysis. Clin Endocrinol (Oxf). 2016; 85: 299–305. https://doi.org/10.1111/cen.13017 PMID: 26732020 7. Noone AM, Cronin KA, Altekruse SF, Howlader N, Lewis DR, Petkov VI, et al. Cancer Incidence and Survival Trends by Subtype Using Data from the Surveillance Epidemiology and End Results Program, 1992–2013. Cancer Epidemiol Biomarkers Prev. 2017; 26: 632–641. https://doi.org/10.1158/1055- 9965.EPI-16-0520 PMID: 27956436 8. Xing M, Alzahrani AS, Carson KA, Viola D, Elisei R, Bendlova B, et al. Association between BRAF V600E mutation and mortality in patients with papillary thyroid cancer. JAMA. 2013; 309: 1493–1501. https://doi.org/10.1001/jama.2013.3190 PMID: 23571588 9. Brown RL, de Souza JA, Cohen EE. Thyroid cancer: burden of illness and management of disease. J Cancer. 2011; 2: 193–199 https://doi.org/10.7150/jca.2.193 PMID: 21509149 10. Mazzaferri EL, Jhiang SM. Long-term impact of initial surgical and medical therapy on papillary and fol- licular thyroid cancer. Am J Med. 1994; 97: 418–428 https://doi.org/10.1016/0002-9343(94)90321-2 PMID: 7977430 11. Xing MZ, Liu RY, Liu XL, Murugan AK, Zhu GW, Zeiger MA, et al. BRAF V600E and TERT Promoter Mutations Cooperatively Identify the Most Aggressive Papillary Thyroid Cancer With Highest Recur- rence. Journal of Clinical Oncology. 2014; 32: 2718–+. https://doi.org/10.1200/JCO.2014.55.5094 PMID: 25024077 PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 10 / 11 V600E Radiomics in predicting BRAF mutation 12. Loh KC, Greenspan FS, Gee L, Miller TR, Yeo PP. Pathological tumor-node-metastasis (pTNM) staging for papillary and follicular thyroid carcinomas: a retrospective analysis of 700 patients. J Clin Endocrinol Metab. 1997; 82: 3553–3562. https://doi.org/10.1210/jcem.82.11.4373 PMID: 9360506 13. Penna GC, Vaisman F, Vaisman M, Sobrinho-Simoes M, Soares P. Molecular Markers Involved in Tumorigenesis of Thyroid Carcinoma: Focus on Aggressive Histotypes. Cytogenet Genome Res. 2016; 150: 194–207. https://doi.org/10.1159/000456576 PMID: 28231576 14. Melo M, Gaspar da Rocha A, Batista R, Vinagre J, Martins MJ, Costa G, et al. TERT, BRAF, and NRAS in Primary Thyroid Cancer and Metastatic Disease. 15. Kim TH, Park YJ, Lim JA, Ahn HY, Lee EK, Lee YJ, et al. The association of the BRAF(V600E) mutation with prognostic factors and poor clinical outcome in papillary thyroid cancer: A Meta-Analysis. Cancer. 2011. https://doi.org/10.1002/cncr.26500 PMID: 21882184 16. Liu R, Bishop J, Zhu G, Zhang T, Ladenson PW, Xing M. Mortality Risk Stratification by Combining BRAF V600E and TERT Promoter Mutations in Papillary Thyroid Cancer: Genetic Duet of BRAF and TERT Promoter Mutations in Thyroid Cancer Mortality. LID - https://doi.org/10.1001/jamaoncol.2016. 3288 PMID: 27581851 17. Gevaert O, Echegaray S, Khuong A, Hoang CD, Shrager JB, Jensen KC, et al. Predictive radioge- nomics modeling of EGFR mutation status in lung cancer. 18. Braman NM, Etesami M, Prasanna P, Dubchuk C, Gilmore H, Tiwari P, et al. Intratumoral and peritu- moral radiomics for the pretreatment prediction of pathological complete response to neoadjuvant che- motherapy based on breast DCE-MRI. 19. Zhang B, Tian J, Dong D, Gu D, Dong Y, Zhang L, et al. Radiomics Features of Multiparametric MRI as Novel Prognostic Factors in Advanced Nasopharyngeal Carcinoma. LID - https://doi.org/10.1158/1078- 0432.CCR-16-2910 PMID: 28280088 20. Yang L, Dong D, Fang M, Zhu Y, Zang Y, Liu Z, et al. Can CT-based radiomics signature predict KRAS/ NRAS/BRAF mutations in colorectal cancer? Eur Radiol. 2018; 28: 2058–2067. https://doi.org/10.1007/ s00330-017-5146-8 PMID: 29335867 21. Li C, Lee KC, Schneider EB, Zeiger MA. BRAF V600E mutation and its association with clinicopatholog- ical features of papillary thyroid cancer: a meta-analysis. J Clin Endocrinol Metab. 2012; 97: 4559– 4570. https://doi.org/10.1210/jc.2012-2104 PMID: 23055546 22. Aerts HJ, Velazquez ER, Leijenaar RT, Parmar C, Grossmann P, Carvalho S, et al. Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach. Nat Commun. 2014; 5: 4006. https://doi.org/10.1038/ncomms5006 PMID: 24892406 23. Harell F. Regression Modeling Strategies With Applicators to Linear Models, Logistic regression, and Survival Analysis. New York, NY: Springer; 2015. 24. Park VY, Han K, Lee E, Kim EK, Moon HJ, Yoon JH, et al. Association Between Radiomics Signature and Disease-Free Survival in Conventional Papillary Thyroid Carcinoma. Sci Rep. 2019; 9: 4501. https://doi.org/10.1038/s41598-018-37748-4 PMID: 30872763 25. Liu T, Zhou S, Yu J, Guo Y, Wang Y, Zhou J, et al. Prediction of Lymph Node Metastasis in Patients With Papillary Thyroid Carcinoma: A Radiomics Method Based on Preoperative Ultrasound Images. Technol Cancer Res Treat. 2019; 18: 1533033819831713. https://doi.org/10.1177/1533033819831713 PMID: 30890092 26. Kim KH, Kang DW, Kim SH, Seong IO, Kang DY. Mutations of the BRAF gene in papillary thyroid carci- noma in a Korean population. Yonsei Med J. 2004; 45: 818–821. https://doi.org/10.3349/ymj.2004.45.5. 818 PMID: 15515191 PLOS ONE | https://doi.org/10.1371/journal.pone.0228968 February 13, 2020 11 / 11

Journal

PLoS ONEPublic Library of Science (PLoS) Journal

Published: Feb 13, 2020

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create folders to
organize your research

Export folders, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off