Contrast-enhanced MRI-based radiomics signatures as new markers for preoperative assessment of the EGFR mutation and subtypes from bone metastases

doi:10.21203/rs.3.rs-1620579/v1

Download PDF

Research Article

Contrast-enhanced MRI-based radiomics signatures as new markers for preoperative assessment of the EGFR mutation and subtypes from bone metastases

https://doi.org/10.21203/rs.3.rs-1620579/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

To develop and externally validate contrast-enhanced (CE) T1-weighted MRI-based radiomics for the identification of epidermal growth factor receptor (EGFR) mutation, exon-19 deletion and exon-21 L858R mutation with the spinal bone metastasis from primary lung adenocarcinoma.

Methods

The study enrolled 159 patients from our hospital between Jan. 2017 and Sep. 2021 to form a primary set, and 24 patients from another center between Jan. 2017 and Oct. 2021 to form an independent validation set. Radiomics features were extracted from the CET1 MRI using the Pyradiomics method. The least absolute shrinkage and selection operator (LASSO) regression was applied for selecting the most predictive features. Radiomics signatures (RSs) were developed based on the primary training set to predict the EGFR mutation and to differentiate the exon-19 deletion and exon-21 L858R. The RSs were validated on the internal and external validation sets using the ROC curve analysis.

Results

A total of 8, 3 and 5 most predictive features were selected to build the RS-EGFR, RS-19 and RS-21 for predicting the EGFR mutation, exon-19 deletion and exon-21 L858R, respectively. The RSs generated favorable prediction efficacies on the primary (AUCs, RS-EGFR vs. RS-19 vs. RS-21, 0.851 vs. 0.816 vs. 0.814) and external validation (AUCs, RS-EGFR vs. RS-19 vs. RS-21, 0.807 vs. 0.742 vs. 0.792) sets.

Conclusions

Radiomics features from the contrast-enhanced MRI could be used to detect the EGFR mutation, increase the certainty of identifying exon-19 deletion and exon-21 L858R mutations based on the spinal metastasis.

EGFR

spinal metastasis

CET1 MRI

lung adenocarcinoma

Lung cancer is the most frequently diagnosed cancer and still increasing in both incidence and mortality worldwide (1–3). Non-small-cell lung cancer (NSCLC) represents approximately 85% of all lung cancer cases, of which lung adenocarcinoma (LUAD) is the most common histologic subtype (4). Identification of epidermal growth factor receptor (EGFR) mutation has an important therapeutic implication in LUAD (5), because tyrosine kinase inhibitors (TKIs) have been reported as an effective therapy for LUAD patients with the EGFR mutation and would lead to satisfactory results (6). Patients harboring EGFR mutations are more sensitive about the EGFR-TKIs compared to those with EGFR wild-type (7). The EGFR mutation mostly occurred in exon 18, 19, 20 and/or 21, among which the 19 deletion and 21 L858R were the most common subtypes (8, 9) and accounted for approximately 90% of EGFR mutation cases (10). Patients who are carrying the EGFR mutation in exon 19 or 21 often have a higher radiographic response rate with EGFR-TKIs (11), a higher response rate to afatinib, erlotinib and gefitinib (11), and hence result longer survival time (12, 13). Therefore, early detection of the EGFR mutation and subtypes are of great significant for therapeutic decisions.

It has been reported that the bone metastasis often occurs in LUAD, and the prevalence is approximately 30% − 40% (14). In clinical, the spinal metastatic lesion would be an important alternative to reflect the EGFR mutation status if the gene test on the primary lesion is unavailable (15). While, the puncture biopsy test can be hardly performed on the spinal metastasis because of potential damages to nerve fibers in the spinal and risks of causing metastases (16, 17).

Magnetic resonance imaging (MRI) is a sensitive method that allows direct visualization of the bone marrow abnormalities (18). The contrast-enhanced (CE) MRI can clearly display the enhanced regions within the tumor, distinguishing necrosis from solid tumors (19). A previous study (20) has shown that the CE MRI is sensitive on the diagnosis of small lesions, which is suitable for bone metastases. Although MRI is a powerful diagnostic technique in NSCLC (21), it is impossible to detect the EGFR mutation in the spinal metastasis by visual examination of the MRI image due to the lack of specific biomarker.

Radiomics is referred to the quantitative analysis of medical imaging, with the capability of digging valuable information from imaging data that can be applied within clinical decision support systems (CDSS) for developing diagnostic and predictive models (22, 23). Many radiomics approaches have been performed in the field of detecting the EGFR mutation in NSCLC. While, the majority of previous works were focused on the EGFR mutations pertaining to the primary lung cancer (24–28). Some recent efforts evaluated associations between radiomics features derived from metastatic lesions in brain and the EGFR mutation status (29–32). Some recent reports revealed that MR imaging features of the bone metastasis were also related to the EGFR mutation status (33, 34). While, the studies only assessed non-contrast enhanced MR data based on limited sample sizes, and lack of external samples to verify their findings, which is inherently limiting. To our knowledge, the relationship between contrast-enhanced MR of bone metastasis and the EGFR mutation status has not yet been clarified. Therefore, the aim of this study is to explore values of contrast-enhanced MR-based radiomics attempting to identify the EGFR mutation and subtypes based on the spinal metastasis.

Patients

This retrospective study was approved by the Medical Ethics Committee of our hospital (20200819YG), and informed consent was waived. A primary set was established with 159 patients enrolled from our hospital between Jan. 2017 and sep.2021. An external set was developed with 24 patients from another hospital between Jan. 2017 and sep.2021. All patients were pathologically diagnosed with the spinal metastasis from original lung adenocarcinoma. The inclusion criteria were: (i) patients older than 18 years; (ii) underwent CET1 MRI scans before surgery; and (iii) patient with complete clinical data. The exclusion criteria were: (i) patients who carry other malignant tumor diseases; (ii) patients who received radiochemotherapy or had phosphate drug; and (iii) patients with vertebral compressed fractures. The primary set was divided into a training set and an internal validation set at a 2:1 ratio by stratified sampling. The external set was used for independent validations. Figure 1 shows the process for the recruitment of patients.

MR image acquisition and metastasis delineation

All MRI scans were performed by using a 3.0 T scanner (Siemens Magnetom Trio, Erlangen, Germany)). The contrast-enhanced T1-weighted MRI parameters were as follows: echo time (TE) = 9.0ms, repetition time (TR) = 550ms, slice thickness = 7 mm, field of view (FOV) = 640×640 mm and matrix size = 256×256. The contrast agent (Gd-DTPA-MBA, Omniscan, GE Healthcare) was injected intravenously at a dose of 0.1 mmol/Kg. Then, 20 mL of saline was flushed at a rate of 3.0 mL/s.

Segmentations of the spinal metastases were manually performed by a radiologist (Xinyan Sun) with 3 years’ experience to generate regions of interest (ROIs) along the border of the metastatic lesions in the CET1 MRI image. The segmentations were cross-checked by a senior radiologist (Yue Dong) with 16 years’ experience. The delineated ROIs were stored in a .NII format.

Feature extraction

The delineated ROIs on the MRI slices were used to extract radiomics features with the pyradiomics package in a Python v.3.6 platform. A set of 1967 features were extracted, which consist first-order statistical, shape-based and textural feature families. Various filters including wavelet, square, squareroot, gradient, exponential, logarithm, localbinarypattern, and laplacian of gaussian were used to transform the original MR images. Then, first-order statistical and textural features were extracted from the transformed images to obtain filtered features. Detailed protocols can be found in a previous study (35) and the pyradiomics document which is available from URL: https://pyradiomics.readthedocs.io/en/latest/.

Feature selection and radiomics model construction

We evaluated all extracted features to select the most predictive features. The intraclass correlation coefficient (ICC) analysis was performed to access interobserver agreements of the features (36). Thirty patients were randomly invited to perform the ICC analysis, 15 of EGFR wild-type and 15 of EGFR mutation. The cut-off value was set as 0.85. Features with ICC > 0.85 were further evaluated with least absolute shrinkage and selection operator (LASSO) regression. The optimal lambda was selected with 5-fold cross-validation. By integrating the selected imaging features with corresponding non-zero LASSO coefficient, the radiomics signature (RS) was developed for predicting the EGFR mutation and subtypes by logistic regression using the “glmnet” package in R v3.6. All models were constructed on the primary training set, and tested on both the internal validation and external validation sets.

Statistical analysis

All statistical analysis were performed in R and MedCalc 20.0.14 (MedCalc Inc.). The two-sided P < 0.05 was considered as significant. Statistical differences in distributions between patients and variables were evaluated with the t-test, Mann-Whitney U test and Chi-Square test as appropriation. The Youden Index (37) was used to decide optimal cutoffs in the ROC analysis. The area under the ROC (AUC), accuracy, specificity (true negative rate) and sensitivity (true positive rate) were calculated to assess the prediction capabilities. Figure 2 showed the workflow of this study.

Patients’ characteristics

All clinical characteristics of the patients were shown in Table 1. There were no significant differences (P > 0.05) in terms of age, gender, smoking status, performance status (PS), carcinoembryonic antigen (CEA), cytokeratin (CYFRA), and neuron-specific enolase (NSE) between the EGFR wild-type and EGFR-mutant groups in the primary and external sets.

Table 1

Clinical characteristics of the patients in the primary and external validation sets.
	Primary training (n = 105)			Internal validation (n = 54)			External validation (n = 24)
Characteristic	EGFR mutant (n = 69)	EGFR wild-type (n = 36)	p	EGFR mutant (n = 35)	EGFR wild-type (n = 19)	p	EGFR mutant (n = 17)	EGFR wild-type (n = 7)	p
Age (Mean ± SD)	58.71 ± 9.34	57.14 ± 11.28	0.448	58.14 ± 12.11	59.21 ± 8.95	0.738	59.06 ± 7.78	60.14 ± 5.61	0.742
Gender			0.080			0.799			0.202
Male	26 (37.7%)	20 (55.6%)		19 (54.3%)	11 (57.9%)		5 (29.4%)	4 (57.1%)
Female	43 (62.3%)	16 (44.4)		16 (45.7%)	8 (42.1%)		12 (70.6%)	3 (42.9%)
Smoking status			0.063			0.532			0.344
Yes	20 (29.0%)	17 (47.25)		10 (28.6%)	7 (36.8%)		4 (23.5%)	3 (42.9%)
No	49 (71.0%)	19 (52.8%)		25 (71.4%)	12 (63.2%)		13 (76.5%)	4 (57.1%)
PS Score			0.553			0.645			0.551
0	8 (11.6%)	2 (5.6%)		3 (8.5%)	1 (5.3%)		3 (17.6%)	3 (42.9%)
1	48 (69.6%)	29 (80.6%)		22 (62.9%)	15 (78.9%)		11 (64.7%)	3 (42.9%)
2	10 (14.5%)	3 (8.3%)		8 (22.9%)	2 (10.5%)		2 (11.8%)	1 (14.2%)
3	3 (4.3%)	2 (5.6%)		2 (5.7%)	1 (5.3%)		1 (5.9%)	0 (0.0%)
CEA (Mean ± SD)	127.01 ± 225.86	118.40 ± 184.17	0.364	84.4 ± 178.92	107.57 ± 276.57	0.738	102 ± 132.94	81.22 ± 124.14	0.187
CYFRA (Mean ± SD)	8.66 ± 13.51	10.51 ± 10.12	0.473	9.8 ± 13.09	12.19 ± 10.34	0.140	14.18 ± 19.25	6.61 ± 3.19	0.852
NSE (Mean ± SD)	21.01 ± 14.59	21.47 ± 11.03	0.787	21.44 ± 16.73	15.22 ± 6.71	0.258	28.11 ± 28.34	18.15 ± 5.99	0.710
SD, standard deviation; PS, performance status; CEA, carcinoembryonic antigen; CYFRA, cytokeratin; NSE, neuron-specific enolase.

Radiomics feature selection

For the prediction of the EGFR mutation, we selected a total of 8 features as the most important predictors. All features belonged to the textural feature family and showed good prediction performance in terms of AUCs. The features showed statistically significant differences (P < 0.05) for predicting the EGFR mutation. For the prediction of the exon-19 deletion and exon-21 L858R mutations, a total of 3 and 5 most important features were selected, respectively. All features generated acceptable predictive performance with AUCs ranging from 0.621 to 0.723. Table 2 listed the prediction performance of the selected features. Figure 3showed correlations among the selected features.

Table 2

Prediction performance of the most important features for predicting the EGFR mutation, exon-19 deletion and exon-21 L858R.
Feature	Mean ± SD				AUC	P
Feature	EGFR mutant	EGFR wild-type	Exon-19 deletion	Exon-21 L858R	AUC	P
exponential_glszm_SmallArea Emphasis (F1)	-0.20 ± 0.87	0.38 ± 1.13	-	-	0.652	0.010
exponential_ngtdm_Strength (F2)	-0.19 ± 0.55	0.36 ± 1.47	-	-	0.629	0.029
log-sigma-3-0-mm-3D_glcm_InverseVariance (F3)	0.17 ± 1.00	-0.32 ± 0.92	-	-	0.649	0.009
log-sigma-5-0-mm-3D_glrlm_Long RunHighGrayLevelEmphasis (F4)	-0.15 ± 0.92	0.29 ± 1.1	-	-	0.640	0.019
wavelet-HHL_glcm_ ClusterShade (F5)	0.11 ± 1.09	-0.2 ± 0.77	-	-	0.632	0.017
wavelet-HHL_glrlm_GrayLevel NonUniformityNormalized (F6)	0.10 ± 1.15	-0.19 ± 0.59	-	-	0.633	0.032
wavelet-HHL_glrlm_Long RunHighGrayLevelEmphasis (F7)	-0.18 ± 0.64	0.34 ± 1.41	-	-	0.624	0.048
wavelet-LLL_glszm_Small AreaLowGrayLevelEmphasis (F8)	-0.03 ± 0.68	0.07 ± 1.44	-	-	0.658	0.011
original_shape_Elongation (F9)	-	-	0.42 ± 0.85	-	0.723	< 0.001
square_glrlm_LongRunHighGray LevelEmphasis (F10)	-	-	0.45 ± 1.40	-	0.675	0.011
wavelet_LLL_firstorder_Skewness (F11)	-	-	0.38 ± 1.27	-	0.647	0.056
lbp-3D-k_glszm_ SmallAreaLowGrayLevelEmphasis (F12)	-	-	-	0.23 ± 0.89	0.628	0.067
log-sigma-3-0-mm-3D_glszm_ SmallAreaLowGrayLevelEmphasis (F13)	-	-	-	0.29 ± 1.34	0.673	0.008
log-sigma-5-0-mm-3D_glcm_Imc2 (F14)	-	-	-	0.33 ± 0.65	0.645	0.031
square_glszm_SizeZoneNonUniformityNormalized (F15)	-	-	-	0.29 ± 0.77	0.621	0.078
wavelet-LHL_firstorder_Median (F16)	-	-	-	0.39 ± 0.66	0.639	0.038

SD, standard deviation

Radiomics signature construction and validation

We used the most important selected features to incorporate with their non-zero coefficients to build RS-EGFR, RS-19 and RS-21 for predicting the EGFR mutation, exon-19 deletion and exon-21 L858R, respectively. The formulas were:

RS-EGFR = -3078.743-4.873929×exponential_glszm_SmallAreaEmphasis-0.9978840× exponential_ngtdm_Strength + 14.92238×log-sigma-3-0-mm-3D_glcm_InverseVariance-0.001164395×log-sigma-5-0-mm-3D_glrlm_LongRunHighGrayLevelEmphasis-117.6612×wavelet-HHL_glcm_ClusterShade-6157.136×wavelet-HHL_glrlm_GrayLevel NonUniformityNormalized-0.1235524×wavelet-HHL_glrlm_Long RunHighGrayLevelEmphasis- 376.3667×wavelet-LLL_glszm_Small AreaLowGrayLevelEmphasis

RS-19 = -1.56251364-2.46002950×original_shape_Elongation + 0.03045322 ×square_glrlm_LongRunHighGrayLevelEmphasis + 1.61180958×wavelet_LLL_firstorder_Skewness

RS-21 = -0.5619225 + 0.9021561×lbp-3D-k_glszm_SmallAreaLowGrayLevelEmphasis + 0.7814898×log-sigma-3-0-mm-3D_glszm_SmallAreaLowGrayLevelEmphasis + 0.7188436×log-sigma-5-0-mm-3D_glcm_Imc2-0.7792080×square_glszm_SizeZoneNonUniformityNormalized + 0.8779048×wavelet-LHL_firstorder_Median

Figure 4 depicted differentiations of the EGFR mutant and wild-type patients by using the developed RSs. The result indicates that our RSs can differentiate among patients with EGFR wild-type, EGFR mutation, exon-19 deletion and exon-21 L858R mutation.

Radiomics model performance

Table 3 listed prediction performance of the RS-EGFR, RS-19 and RS-21 in the primary and external validation sets. ROC curves of the RSs were shown in Fig. 5. For the prediction of EGFR mutation, the RS-EGFR generated AUCs of 0.851 and 0.807 in the primary and external validation sets, respectively. For prediction of the exon-19 deletion and exon-21 L858R mutations, the RS-19 and RS-21 yielded predictive AUCs of 0.816 and 0.742, 0.821 and 0.764 in the primary and external validation sets, respectively

Table 3

Prediction performance of RS-EGFR, RS-19 and RS-21.
RS	Primary training				Internal validation				External validation
RS	AUC (95%)	ACC	SEN	SPE	AUC (95%)	ACC	SEN	SPE	AUC (95%)	ACC	SEN	SPE
RS-EGFR	0.851 (0.774–0.921)	0.800	0.722	0.870	0.780 (0.645–0.916)	0.741	0.771	0.737	0.807 (0.595–0.938)	0.625	0.765	0.857
RS-19	0.816 (0.716–0.917)	0.739	0.964	0.561	0.789 (0.636–0.942)	0.600	0.786	0.714	0.742 (0.478–0.919)	0.623	0.833	0.636
RS-21	0.814 (0.714–0.914)	0.739	0.655	0.875	0.770 (0.609–0.931)	0.657	0.800	0.700	0.792 (0.530–0.946)	0.529	0.750	0.889

AUC: area under the receiver operating characteristic curve; ACC: accuracy; SEN: sensitivity; SPE: specificity.

Non-invasive evaluation of the EGFR mutation status and subtypes based on the bone metastasis are of great clinical significance, yet not being well studied. Previous studies related to our work mainly focused on the primary lesions of lung adenocarcinoma, and highlighted that radiomics can be helpful to predict the EGFR mutation and subtypes mainly from medical imaging (10, 21, 26, 38, 39). In recent years, the contrast-enhanced MRI has been demonstrated to provide information about the vascularity of tissue and its surrounding environment (10), and to detect the increase in blood vessels of the tumor in the paraspinal muscles and spinal canal, thereby assessing the spread of the tumor outside the skeletal system (40). To our knowledge, this study was the first attempt to investigate values of contrast-enhanced MRI on the assessment of the EGFR mutation status in the bone metastasis.

A total of 1967 radiomics features were calculated and evaluated from the MRI image. For predicting EGFR mutation, a total of 8 features were identified as the most predictive one, all belongs to the textural features. This may suggest that the intratumor heterogeneity within the spinal metastasis was related to the EGFR mutation status, considering the textural features can reflect the intratumoral non-uniformity(41). This was partly consistent with Lindberg’s work that demonstrated the relationship between the EGFR mutation and the heterogeneity, and aggressiveness of lung cancer (42). For prediction of the exon-19 deletion and exon-21 L858R, we identified 3 and 5 most important features from the CET1 MRI, respectively. Notably, most of the selected features (4 of 5) for predicting the exon-21 L858R belong to the textural feature class. This may indicate that heterogeneity distributions within the metastasis were different between the tumor carrying exon-19 deletion and exon-21 L858R mutations. The selected original_shape_elongation feature describes shapes of the ROI. The higher value of this feature indicates the rounder shape of the tumor. Our findings may suggest that the tumor harboring exon-19 deletion was tended to be rounder compared with those carrying exon-21 L858R or EGFR wild-type.

Comparing with Jiang’s work (33), our proposed CET1 MRI-based RS-EGFR generated better AUCs than non-contrast enhanced sequences. This may indicate that the CE MRI contains more predictive information related to the EGFR status. In addition, the main advantage of this study is the comprehensive analysis of the two common subtypes. Before this study, investigations on predicting the exon-19 deletion and exon-21 L858R were limited. Li et al (10) generated an AUC of 0.79 for detecting the exon-19 deletion and exon-21 L858R using radiomics based on primary lung adenocarcinoma. Cao et al (43) yielded an AUC of 0.901 for detecting the exon-19 deletion and exon-21 L858R based on the metastasis from original lung adenocarcinoma, but they used a small sample size and was lack of an external validation set, which was inherently limiting. In this study, we first predicted the EGFR mutation, then further assessed whether the mutation was located in exon 19 or 21. Our results revealed that features from contrast-enhanced MRI have good potentials to detect the EGFR mutation and subtypes.

There were controversies about the relationship between clinical factors and the EGFR mutation and subtypes. The age and smoking were demonstrated to be associated with the EGFR mutation (26, 40, 44) ad subtypes (10), but pointed to be not predictive in some recent studies(34, 39). Besides, the CEA level and gender were also controversial. The two factors were suggested as independent predictors in previous works (25, 45), but found to be ineffective in another reports (34, 46). In our study, no clinical factor was found to be related to the EGFR mutation and subtypes in both the primary and external cohorts, which may be due to the limited number of patients.

There are limitations in this study. First, although the findings were externally validated on an independent set, the sample size was still small, which was due to data collection challenges. The identified most important features from the CET1 MRI might be verified on a large scale. Second, some important MRI parameters were not included in the developed radiomics models. Third, we did not assess the T790M mutation that is also important in improving treatment managements.

In conclusion, this study assessed values of CET1 MR on prediction of the EGFR mutation and subtypes based on the spinal metastasis. The developed radiomics models performed well on the external set, which may indicate good potential in future clinical applications.

AUC

Area under the Curve

ROC

Receiver operating characteristic

EGFR-TKIs

Epidermal growth factor receptor-tyrosine kinase inhibitors

NSCLC

Non-small-cell lung carcinoma

MRI

Magnetic resonance imaging

LASSO

Least absolute shrinkage and selection operator

AIC

Akaike information criterion

EGFR

Epidermal growth factor receptor

CET1

Contrast-enhanced T1-weighted imaging

ICC

Intraclass correlation coefficient (ICC)

Performance status

CEA

Carcinoembryonic antigen

CYFRA

Cytokeratin NSE:Neuron-specific enolase

LUAD

Lung adenocarcinoma

Acknowledgments

Not applicable.

Authors' contributions

Ying Fan and Yue Dong contributed to the conception and design of the study. Xiran Jiang, Hongbo Wang and Xinyan Sun contributed to the acquisition of the data. Ying Fan, Yue Dong and Xiran Jiang contributed to the data analysis and interpretation. Huan Wang, Xinyan Sun, Peng Zhao contributed to statistical analysis of this study. Xiran Jiang and Ying Fan participated in Manuscript preparation and manuscript editing of this study. Xiran Jiang and Yue Dong participated in the review of manuscript.

Funding

The study was funded by Climbing Fund of National Cancer Center (NCC201806B011), National Natural Science Foundation of China (81872363), PhD Start-up Fund of Liaoning Province (2021-BS-044), China National Natural Science Foundation (31770147), Natural Science Foundation of Liaoning Province (2021-MS-205) and Medical-Engineering Joint Fund for Cancer Hospital of China Medical University and Dalian University of technology (LD202029).

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethical approval and consent to participate

All analyses of human data conducted in this study were approved by the Institutional Review Board of the Cancer Hospital of China Medical University and in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was exempted because of the retrospective nature of this study. The requirement for informed consent was waived by the Ethics Committee of (Liaoning Cancer Hospital and Institute) because of the retrospective nature of the study (20200819YG).

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Author’s details

¹ Department of Biomedical Engineering, School of Intelligent Medicine, China Medical University, Shenyang, 110122, P.R. China ² Department of Biophysics, School of Intelligent Medicine, China Medical University, Shenyang, 110122, P.R. China³ Department of Radiology, Cancer Hospital of China Medical University, Liaoning, 110042, P.R. China ⁴Department of Radiology, Shengjing Hospital of China Medical University, Shenyang, 110004, P.R. China ⁵ Department of Scientific Research and Academic, Cancer Hospital of China Medical University, Liaoning Cancer Hospital and Institute, Liaoning, 110042, P.R. China

Gridelli C, Rossi A, Carbone DP, et al. Non-small-cell lung cancer. Nat Rev Dis Primers 2015;1:15009.
Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, et al. Cancer incidence and mortality patterns in Europe: estimates for 40 countries in 2012. Eur J Cancer 2013;49:1374–403.
Torre LA, Bray F, Siegel RL, et al. Global cancer statistics, 2012. CA Cancer J Clin 2015;65:87–108.
Travis WD, Brambilla E, Riely GJ. New pathologic classification of lung cancer: relevance for clinical practice and clinical trials. J Clin Oncol 2013;31:992–1001.
da Cunha Santos G, Shepherd FA, Tsao MS. EGFR mutations and lung cancer. Annu Rev Pathol 2011;6:49–69.
Hu F, Li C, Xu J, et al. Additional local consolidative therapy has survival benefit over EGFR tyrosine kinase inhibitors alone in bone oligometastatic lung adenocarcinoma patients. Lung Cancer 2019;135:138–44.
Recondo G, Facchinetti F, Olaussen KA, et al. Making the first move in EGFR-driven or ALK-driven NSCLC: first-generation or next-generation TKI? Nature Reviews Clinical Oncology 2018;15:694–708.
Yano M, Sasaki H, Kobayashi Y, et al. Epidermal growth factor receptor gene mutation and computed tomographic findings in peripheral pulmonary adenocarcinoma. J Thorac Oncol 2006;1:413–6.
Locatelli-Sanchez M, Couraud S, Arpin D, et al. Routine EGFR molecular analysis in non-small-cell lung cancer patients is feasible: exons 18–21 sequencing results of 753 patients and subsequent clinical outcomes. Lung 2013;191:491–9.
Li S, Ding C, Zhang H, et al. Radiomics for the prediction of EGFR mutation subtypes in non-small cell lung cancer. Med Phys 2019;46:4545–52.
Carey KD, Garton AJ, Romero MS, et al. Kinetic analysis of epidermal growth factor receptor somatic mutant proteins shows increased sensitivity to the epidermal growth factor receptor tyrosine kinase inhibitor, erlotinib. Cancer Res 2006;66:8163–71.
Fukuoka M, Wu YL, Thongprasert S, et al. Biomarker analyses and final overall survival results from a phase III, randomized, open-label, first-line study of gefitinib versus carboplatin/paclitaxel in clinically selected patients with advanced non-small-cell lung cancer in Asia (IPASS). J Clin Oncol 2011;29:2866–74.
Zheng H, Zhang Y, Zhan Y, et al. Prognostic analysis of patients with mutant and wild-type EGFR gene lung adenocarcinoma. Cancer Manag Res 2019;11:6139–50.
Kuchuk M, Kuchuk I, Sabri E, et al. The incidence and clinical impact of bone metastases in non-small cell lung cancer. Lung Cancer 2015;89:197–202.
Krawczyk P, Nicos M, Ramlau R, et al. The Incidence of EGFR-Activating Mutations in Bone Metastases of Lung Adenocarcinoma. Pathology & Oncology Research 2014;20:107–12.
Shen TX, Liu L, Li WH, et al. CT imaging-based histogram features for prediction of EGFR mutation status of bone metastases in patients with primary lung adenocarcinoma. Cancer Imaging 2019;19:34.
Song J, Shi J, Dong D, et al. A New Approach to Predict Progression-free Survival in Stage IV EGFR-mutant NSCLC Patients with EGFR-TKI Therapy. Clin Cancer Res 2018;24:3583–92.
Hanna SL, Fletcher BD, Fairclough DL, et al. Magnetic resonance imaging of disseminated bone marrow disease in patients treated for malignancy. Skeletal Radiol 1991;20:79–84.
Zhou M, Scott J, Chaudhury B, et al. Radiomics in Brain Tumor: Image Assessment, Quantitative Feature Descriptors, and Machine-Learning Approaches. AJNR Am J Neuroradiol 2018;39:208–16.
Russell EJ, Geremia GK, Johnson CE, et al. Multiple cerebral metastases: detectability with Gd-DTPA-enhanced MR imaging. Radiology 1987;165:609–17.
Wang S, Shi J, Ye Z, et al. Predicting EGFR mutation status in lung adenocarcinoma on computed tomography image using deep learning. European Respiratory Journal 2019;53.
Guiot J, Vaidyanathan A, Deprez L, et al. A review in radiomics: Making personalized medicine a reality via routine imaging. Med Res Rev 2021.
Lambin P, Leijenaar RTH, Deist TM, et al. Radiomics: the bridge between medical imaging and personalized medicine. Nat Rev Clin Oncol 2017;14:749–62.
Jia TY, Xiong JF, Li XY, et al. Identifying EGFR mutations in lung adenocarcinoma by noninvasive imaging using radiomics features and random forest modeling. Eur Radiol 2019;29:4742–50.
Tu W, Sun G, Fan L, et al. Radiomics signature: A potential and incremental predictor for EGFR mutation status in NSCLC patients, comparison with CT morphology. Lung Cancer 2019;132:28–35.
Hong D, Xu K, Zhang L, et al. Radiomics Signature as a Predictive Factor for EGFR Mutations in Advanced Lung Adenocarcinoma. Frontiers in Oncology 2020;10.
Wu S, Shen G, Mao J, et al. CT Radiomics in Predicting EGFR Mutation in Non-small Cell Lung Cancer: A Single Institutional Study. Frontiers in Oncology 2020;10.
Liu Y, Kim J, Balagurunathan Y, et al. Radiomic Features Are Associated With EGFR Mutation Status in Lung Adenocarcinomas. Clinical Lung Cancer 2016;17:441–8.
Ahn SJ, Kwon H, Yang J-J, et al. Contrast-enhanced T1-weighted image radiomics of brain metastases may predict EGFR mutation status in primary lung cancer. Scientific Reports 2020;10.
Chen BT, Jin T, Ye N, et al. Radiomic prediction of mutation status based on MR imaging of lung cancer brain metastases. Magn Reson Imaging 2020;69:49–56.
Wang GY, Wang BM, Wang Z, et al. Radiomics signature of brain metastasis: prediction of EGFR mutation status. European Radiology 2021;31:4538–47.
Park YW, Ahn SS, Choi D, et al. CMET-04. Radiomics features can differentiate the EGFR mutation status of brain metastases from non-small cell lung cancer. 2019;21:vi51.
Jiang X, Ren M, Shuang X, et al. Multiparametric MRI-Based Radiomics Approaches for Preoperative Prediction of EGFR Mutation Status in Spinal Bone Metastases in Patients with Lung Adenocarcinoma. Journal of Magnetic Resonance Imaging 2021;54:497–507.
Ren M, Yang H, Lai Q, et al. MRI-based radiomics analysis for predicting the EGFR mutation based on thoracic spinal metastases in lung adenocarcinoma patients. Med Phys 2021;48:5142–51.
van Griethuysen JJM, Fedorov A, Parmar C, et al. Computational Radiomics System to Decode the Radiographic Phenotype. Cancer Research 2017;77:E104-E7.
Li Z, Duan H, Zhao K, et al. Stability of MRI radiomics features of hippocampus: an integrated analysis of test-retest and inter-observer variability. 2019;7:97106–16.
Ruopp MD, Perkins NJ, Whitcomb BW, et al. Youden index and optimal cut-point estimated from observations affected by a lower limit of detection. Biometrical Journal 2008;50:419–30.
Yuan M, Pu X-H, Xu X-Q, et al. Lung adenocarcinoma: Assessment of epidermal growth factor receptor mutation status based on extended models of diffusion-weighted image. Journal of Magnetic Resonance Imaging 2017;46:281–9.
Liu G, Xu Z, Ge Y, et al. 3D radiomics predicts EGFR mutation, exon-19 deletion and exon-21 L858R mutation in lung adenocarcinoma. Translational Lung Cancer Research 2020;9:1212–24.
Mei D, Luo Y, Wang Y, et al. CT texture analysis of lung adenocarcinoma: can Radiomic features be surrogate biomarkers for EGFR mutation statuses. Cancer Imaging 2018;18:52.
Mohammadi A, Afshar P, Asif A, et al. Lung Cancer Radiomics: Highlights from the IEEE Video and Image Processing Cup 2018 Student Competition. IEEE Signal Process Mag 2019;36:164–73.
Lindberg OR, McKinney A, Engler JR, et al. GBM heterogeneity as a function of variable epidermal growth factor receptor variant III activity. Oncotarget 2016;7:79087–102.
Cao R, Dong Y, Wang X, et al. MRI-Based Radiomics Nomogram as a Potential Biomarker to Predict the EGFR Mutations in Exon 19 and 21 Based on Thoracic Spinal Metastases in Lung Adenocarcinoma. Acad Radiol 2021.
Zhang J, Zhao X, Zhao Y, et al. Value of pre-therapy F-18-FDG PET/CT radiomics in predicting EGFR mutation status in patients with non-small cell lung cancer. European Journal of Nuclear Medicine and Molecular Imaging 2020;47:1137–46.
Ko K-H, Hsu H-H, Huang T-W, et al. Value of F-18-FDG uptake on PET/CT and CEA level to predict epidermal growth factor receptor mutations in pulmonary adenocarcinoma. European Journal of Nuclear Medicine and Molecular Imaging 2014;41:1889–97.
Zhang J, Zhao X, Zhao Y, et al. Value of pre-therapy (18)F-FDG PET/CT radiomics in predicting EGFR mutation status in patients with non-small cell lung cancer. Eur J Nucl Med Mol Imaging 2020;47:1137–46.

No competing interests reported.

Download PDF

Editorial decision: Major revision
25 Jul, 2022
Reviews received at journal
10 Jul, 2022
Reviewers agreed at journal
28 Jun, 2022
Reviewers agreed at journal
19 Jun, 2022
Reviewers invited by journal
04 Jun, 2022
Editor assigned by journal
04 Jun, 2022
Editor invited by journal
13 May, 2022
Submission checks completed at journal
12 May, 2022
First submitted to journal
03 May, 2022

You are reading this latest preprint version

Contrast-enhanced MRI-based radiomics signatures as new markers for preoperative assessment of the EGFR mutation and subtypes from bone metastases

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Introduction

Methods

Patients

MR image acquisition and metastasis delineation

Feature extraction

Feature selection and radiomics model construction

Statistical analysis

Results

Patients’ characteristics

Radiomics feature selection

Radiomics signature construction and validation

Radiomics model performance

Discussion

Limitation

Conclusion

Abbreviations

Declarations

References

Additional Declarations

Status:

Version 1