Integrating 18 F-FDG PET/CT Radiomics and Body Composition for Enhanced Prognostic Assessment in Patients with Esophageal Cancer

doi:10.21203/rs.3.rs-4827934/v1

Download PDF

Research Article

Integrating 18 F-FDG PET/CT Radiomics and Body Composition for Enhanced Prognostic Assessment in Patients with Esophageal Cancer

https://doi.org/10.21203/rs.3.rs-4827934/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

This study aimed to develop a predictive model utilizing radiomics and body composition features derived from ¹⁸F-FDG PET/CT scans to forecast progression-free survival (PFS) and overall survival (OS) outcomes in patients with esophageal squamous cell carcinoma (ESCC).

Methods

We analyzed data from 91 patients who underwent baseline ¹⁸F-FDG PET/CT imaging. Radiomic features extracted from PET and CT images and subsequent radiomics scores (Rad-scores) were calculated. Body composition metrics were also quantified, including muscle and fat distribution at the L3 level from CT scans. Multiparametric survival models were constructed using Cox regression analysis, and their performance was assessed using the area under the time-dependent receiver operating characteristic (ROC) curve (AUC) and concordance index (C-index).

Results

Multivariate analysis identified Rad-score_PFS (P = 0.003), sarcopenia (P < 0.001), and visceral adipose tissue index (VATI) (P < 0.001) as independent predictors of PFS. For OS, Rad-score_OS (P = 0.001), sarcopenia (P = 0.002), VATI (P = 0.037), stage (P = 0.042), and body mass index (BMI) (P = 0.008) were confirmed as independent prognostic factors. Integration of the Rad-score with clinical variables and body composition parameters enhanced predictive accuracy, yielding C-indices of 0.810 (95% CI: 0.737–0.884) for PFS and 0.806 (95% CI: 0.720–0.891) for OS.

Conclusions

This study underscored the potential of combining Rad-score with clinical and body composition data to refine prognostic assessment in ESCC patients.

Esophageal squamous cell carcinoma

Radiomics

Body composition

PET/CT

Prognosis

Esophageal cancer (EC) ranks as the seventh most prevalent cancer and the sixth leading cause of cancer-related mortality globally, with esophageal squamous cell carcinoma (ESCC) constituting the predominant histological subtype ^[1]. Prognosis in ESCC heavily hinges on tumor stage; patients diagnosed at an early stage or with precancerous lesions exhibit a relatively high 5-year survival rate, contrasting sharply with those presenting advanced ESCC, where survival rates drop below 30% ^[2–4]. Therefore, timely identification of patients with adverse clinical features and effective intervention strategies are critical for enhancing prognosis.

Research indicates a significant correlation between body composition, particularly muscle and adipose tissue (AT), and prognosis across various malignancies, including EC ^[5–8]. Malnutrition affects approximately 60–80% of EC patients due to symptoms such as obstruction, nausea, and dysphagia associated with the disease ^[9], correlating with poorer clinical outcomes ^[10]. Sarcopenia, characterized by skeletal muscle mass and function loss often accompanied by AT accumulation (sarcopenic obesity), represents a prevalent adverse effect of cancer and its treatments, with prevalence rates ranging from 14–79% depending on age, malnutrition status, and cancer type ^{[11, 12]}. Moreover, the composition of AT, particularly visceral fat, emerges as a crucial factor in cancer prognosis research ^[13].

Radiomics represents a reproducible, non-invasive tool that extracts quantitative features from pre-treatment biomedical images (e.g., CT, MRI, and PET), providing valuable supplementary information for predictive modeling ^{[14, 15]}. These features can be utilized independently or combined with clinical and pathological data to develop novel biomarker combinations aimed at enhancing prognostic accuracy ^[16]. Several studies have demonstrated promising outcomes by integrating independent biomarkers such as clinical risk factors and radiomics ^[17–20].

This study aimed to construct a predictive model that integrated radiomic features and body composition data obtained from PET/CT scans to identify high-risk individuals among patients with squamous ESCC.

2.1. Patient data

Patients diagnosed with primary ESCC and who underwent ¹⁸F-FDG PET/CT imaging at our institution between April 2013 and December 2022 as part of their initial staging were included in this retrospective study. Approval from the institutional research ethics board was obtained, and the requirement for informed consent was waived.

2.2. Image acquisition

All patients fasted for at least 6 hours prior to imaging to ensure blood glucose levels were maintained below 11 mmol/L. Each patient received an intravenous injection of 3.5–4.5 MBq/kg of FDG. PET/CT scans were conducted using a GE Discovery STE PET/CT scanner, capturing images from the base of the skull to the upper thigh approximately 60 minutes post-injection of the radiotracer.

2.3. Body composition measurements

Measurements of skeletal muscle index (SMI), subcutaneous adipose tissue index (SATI), and visceral adipose tissue index (VATI) were performed by analyzing the L3 vertebral level on CT images obtained from PET/CT scans using slice-O-Matic software (version 5.0; TomoVision) (Fig. 1) ^[21]. Hounsfield unit (HU) thresholds were set at − 29 to + 150 HU for muscle area, − 190 to − 30 HU for subcutaneous fat, and − 150 to − 50 HU for peri-visceral fat ^[22]. Body mass index (BMI) was calculated as weight (kg) divided by squared height (m²). Skeletal muscle and AT areas were normalized to height squared (m²) to derive SMI, VATI, and SATI. Sarcopenia was defined as SMI < 34.4 cm²/m² in females and < 45.4 cm²/m² in males ^[23].

2.4. Tumor segmentation and feature extraction

Each PET/CT scan included segmentation of all ¹⁸F-FDG–avid primary tumor lesions, independently performed by two PET/CT diagnostic physicians using a semi-automatic approach with LIFEx software (version 6.30; https://www.lifexsoft.org/) ^[24], applying a threshold set at 40% of SUV_max. Prior to texture feature extraction, PET and CT images underwent preprocessing steps, which included: (1) spatial resampling to achieve voxel dimensions of 2 × 2 × 2 mm³; (2) intensity discretization of PET data into 64 gray levels with scale bounds ranging from 0 to 20; and (3) intensity rescaling of CT data using 400 gray levels within the range of -1,000 to 3,000 HU. Radiomics features were simultaneously extracted from PET and CT images within the defined volume of interest (VOI). A total of 92 radiomic features were extracted, encompassing conventional indexes, first-order histogram indexes, shape indexes, gray-level co-occurrence matrices (GLCM), gray-level run-length matrices (GLRLM), neighboring gray-level difference matrices (NGLDM), and gray-level zone length matrices (GLZLM).

2.5. Feature selection and Rad-score construction

To ensure the reproducibility and reliability of radiomics features, only features with intraclass correlation coefficients (ICCs) greater than 0.75 were selected for subsequent analysis. The least absolute shrinkage and selection operator (LASSO) method was then employed to identify prognostically associated features. Subsequently, the radiomics score (Rad-score) for each patient was computed as a linear combination of these features with corresponding coefficients. Prognostic factors were further selected using univariate and multivariate Cox regression analyses, considering P-values less than 0.05. The workflow of the study is illustrated in Fig. 2.

2.6. Statistical analysis

Statistical analyses were performed using R software version 4.3.1 and SPSS version 26.0. A significance level of P ≤ 0.05 was considered statistically significant. Baseline characteristics were compared using Student’s t-test, Mann-Whitney U-test, chi-squared test, and Fisher’s exact test as appropriate. Cox regression analysis was employed to identify potential prognostic factors and construct predictive models. Progression-free survival (PFS) and overall survival (OS) were evaluated using the Kaplan-Meier method and compared between groups using the log-rank test. Additionally, a nomogram was developed to visualize the outcomes predicted by the final model. The performance of the model was assessed using the area under the time-dependent receiver operating characteristic (ROC) curve (AUC) and concordance index (C-index).

3.1. Patient characteristics

A total of 91 patients diagnosed with ESCC were included in the study, comprising 69 men and 22 women. The median age of the cohort was 67 years (range: 43–85 years). All patients underwent baseline ¹⁸F-FDG PET/CT for staging purposes. Detailed baseline characteristics are summarized in Table 1.

Table 1

Patients’ characteristics
Characteristic	Patients (n = 91)
Age (y)	67 (43–85)
Sex
M	69 (75.82%)
F	22 (24.18%)
Tumor stage
Stage I-II	37 (40.66%)
Stage III-IV	54 (59.34%)
LVI
Yes	45 (49.45%)
No	46 (50.55%)
Number of PET-positive LNs ≥ 2	39 (42.86%)
Treatment
Surgery alone	16 (17.58%)
Neo and adjuvant treatment + surgery	28 (30.77%)
Radiochemotherapy (RCT)	22 (24.18%)
Chemotherapy alone	15 (16.48)
Radiotherapy alone	3 (3.30%)
Lost to follow-up	7 (7.69%)
BMI, Kg/m²	21.61 (16.00-31.60)
BMI category
< 18.5 kg/m², underweight	19
18.5–24.9 kg/m², normal weight	55
25-29.9 kg/m², overweight	14
≥ 30 kg/m², obese	3
SMI (cm²/m²)	43.16 (25.41–67.69)
SATI (cm²/m²)	27.18 (0.12–84.76)
VATI (cm²/m²)	29.18 (1.41–93.26)
Sarcopenia
Yes	42 (46.2%)
No	49 (53.8%)
Rad-score_PFS	0.498 (0.183–0.984)
Rad-score_OS	0.473 (0.254–0.733)
Abbreviations: M, male; F, female; LVI, lymphovascular invasion; BMI, body mass index; SMI, skeletal muscle index; SATI, subcutaneous adipose tissue index; VATI, visceral adipose tissue index.

3.2. Factors associated with body composition index

Table 2 displays the characteristics of patients diagnosed with sarcopenia (low SMI), totaling 42 individuals. Patients with low SMI exhibited a higher prevalence of PET-positive lymph nodes (59.0% [23 out of 91 patients] vs. 41.0% [16 out of 91 patients], P = 0.034), lower BMI (19.59 kg/m² vs. 23.66 kg/m²; P < 0.001), reduced VATI (15.00 cm²/m² vs. 36.82 cm²/m²; P < 0.001), and decreased SATI (18.38 cm²/m² vs. 37.32 cm²/m²; P < 0.001).

Table 2

Patient grouping based on body composition indexes
	SMI			SATI			VATI
Variable	Sarcopenia (n = 42)	Non-sarcopenia (n = 49)	P Value	High (n = 38)	Low (n = 53)	P value	High (n = 83)	Low (n = 8)	P value
Age (y)	67 (49–82)	67 (43–85)	0.416	69 (52–82)	66 (43–85)	0.058	67(43–84)	68.5 (52–85)	0.074
Sex			0.121			< 0.001*			0.674
M	35 (50.7)	34 (49.3)		21 (30.4)	48 (69.6)		62(89.9)	7(10.1)
F	7 (31.8)	15(68.2)		17 (77.3)	5 (22.7)		21(95.5)	1(4.5)
Tumor stage			0.374			0.270			0.465
Stage I-II	15 (40.5)	22 (59.5)		18 (48.6)	19 (51.4)		35 (94.6)	2 (5.4)
Stage III-IV	27 (50.0)	27 (50.0)		20 (37.0)	34 (63.0)		48 (88.9)	6 (11.1)
LVI			0.174			0.235			0.030*
Yes	24 (53.3)	21(46.7)		16 (35.6)	29 (64.4)		38 (84.4)	7 (15.6)
No	18 (39.1)	28 (60.9)		22 (47.8)	24 (52.2)		45 (97.8)	1 (2.2)
BMI	19.59(16.00-27.55)	23.66(17.58–31.60)	< 0.001*	23.83(17.58–30.43)	20.44(16.00-31.40)	< 0.001*	21.77(16.36–31.60)	17.95 (16.00-22.72)	< 0.001*
BMI			< 0.001*			0.001*			0.001*
< 18.5 kg/m², underweight	17 (89.5)	2 (10.5)		3 (15.8)	16 (84.2)		13 (68.4)	6(31.6)
18.5–24.9 kg/m², normal weight	22 (40.0)	33 (60.0)		21 (38.2)	34 (61.8)		53 (96.4)	2(3.6)
25-29.9 kg/m², overweight	3 (21.4)	11 (78.6)		12 (85.7)	2 (14.3)		14 (100.0)	0 (0.0)
≥ 30 kg/m², obese	0(0.0)	3 (100.0)		2 (66.7)	1 (33.3)		3 (100.0)	0(0.0)
Number of PET-positive LNs			0.034*			0.326			0.281
0–1	19 (36.5)	33 (63.5)		24 (46.2)	28 (53.8)		49 (94.2)	3 (5.8)
≥ 2	23 (59.0)	16(41.0)		14 (35.9)	25 (64.1)		34 (87.2)	5(12.8)
SMI				45.35(25.41–67.69)	41.59 (25.85–60.38)	0.034*	43.77 (25.41–67.69)	35.43(27.16–44.99)	0.015*
SMI			-			< 0.001*			0.001*
Yes	-	-		6 (14.3)	36 (85.7)		34 (81.0)	8 (19.0)
No	-	-		32 (65.3)	17 (34.7)		49 (100.0)	0 (0.0)
SATI	18.38(0.12–61.38)	37.32(7.00-84.76)	< 0.001*				29.76(2.41–84.76)	2.49(0.12–9.24)	< 0.001*
VATI	15.00(1.41–93.26)	36.82(6.45–83.27)	< 0.001*	44.02(16.58–93.26)	14.52(1.41–48.98)	0.047*
Rad-score_PFS	0.51(0.18–0.983)	0.49(0.20-.97)	0.133	0.50(0.24–0.97)	0.50(0.18–0.98)	0.485	0.50(0.20–0.98)	0.51(0.18–0.61)	0.785
Rad-score_OS	0.48(0.37–0.73)	0.46(0.25–0.62)	0.527	0.46(0.36–0.60)	0.49(0.25–0.73)	0.061	0.46(0.25–0.73)	0.53(0.37–0.65)	0.186
Abbreviations: M, male; F, female; LVI, lymphovascular invasion; BMI, body mass index; SMI, skeletal muscle index; SATI, subcutaneous adipose tissue index; VATI, visceral adipose tissue index.
*P < 0.05

The survival ROC analysis determined optimal cut-off values for body composition. Among the patients, 53 had low SATI and 38 had high SATI, while eight had low VATI and 83 had high VATI. Compared to those with high SATI, patients with low SATI were predominantly male (69.6% [48 out of 91 patients] vs. 30.4% [21 out of 91 patients], P < 0.001), had lower BMI (20.44 kg/m² vs. 23.83 kg/m², P < 0.001), reduced SMI (41.59 cm²/m² vs. 45.35 cm²/m²; P = 0.034), and lower VATI (14.52 cm²/m² vs. 44.02 cm²/m²; P = 0.047). Similarly, patients in the low VATI group had significantly lower BMI (17.95 kg/m² vs. 21.77 kg/m²; P < 0.001), reduced SMI (35.43 cm²/m² vs. 43.77 cm²/m²; P = 0.015), and lower SATI (2.49 cm²/m² vs. 29.76 cm²/m²; P < 0.001).

3.3. Prognostic impact

The median follow-up time for the entire cohort was 18 months (range: 1–69 months). Among the patients, 38 (41.76%) experienced disease progression or relapse, with a median time to progression of 15 months, while 28 (30.77%) patients died, with a median survival time of 18 months. The 1-year, 2-year, and 5-year PFS rates were 76.92%, 68.13%, and 58.24%, respectively, and the corresponding OS rates were 84.62%, 75.82%, and 69.23%.

Univariable Cox regression analysis identified staging, lymphovascular invasion (LVI), BMI, number of PET-positive lymph nodes, Rad-score, VATI, and sarcopenia as significant prognostic factors for both PFS and OS. In multivariate analysis, Rad-score_PFS (P = 0.003), sarcopenia (P < 0.001), and VATI (P < 0.001) remained independent predictors. For OS, Rad-score_OS (P = 0.001), sarcopenia (P = 0.002), VATI (P = 0.037), stage (P = 0.042), and BMI (P = 0.008) were confirmed as independent prognostic factors (Table 3 and Fig. 3).

Table 3

Cox proportional hazard regression analysis for PFS and OS
Variable	5-PFS			5-OS
	P value	HR	95%CI	P value	HR	95%CI
Univariate
Age	0.168	0.630	0.327–1.216	0.282	0.658	0.307–1.411
Sex	0.584	0.799	0.358–1.784	0.312	0.621	0.247–1.562
Stage (III-IV)	0.027*	2.072	1.085–3.959	0.026*	2.351	1.108–4.988
LVI	0.011*	2.330	1.216–4.466	0.037*	2.218	1.048–4.693
BMI	0.035*	1.284	0.434–3.796	0.003*	0.657	0.190–2.276
Number of PET-positive LNs (≥ 2)	0.010*	2.404	1.238–4.670	0.029*	2.330	1.090–4.980
Rad-score	< 0.001*	4.705	2.219–9.979	< 0.001*	4.7053	2.219–9.979
Sarcopenia	0.004*	2.613	1.364–5.003	0.003*	3.144	1.480–6.677
SATI	0.308	1.594	0.651–3.908	0.399	0.7206	0.336–1.544
VATI	0.026*	2.123	1.095–4.115	< 0.001*	0.0503	0.010–0.254
Multivariate
Stage (III-IV)	0.189	2.516	0.636–9.957	0.042*	4.984	1.060-23.445
LVI	0.376	1.846	0.475–7.177	0.107	0.281	0.060–1.313
BMI	0.083	1.608	0.940–2.752	0.008*	2.220	1.233–3.997
Rad-score	0.003*	4.000	1.600–10.000	0.001*	6.865	2.148–21.947
Sarcopenia	< 0.001*	5.742	2.349–14.037	0.002*	4.477	1.704–11.759
VATI	< 0.001*	6.663	2.806–15.821	0.037*	0.323	0.111–0.935
Abbreviations: LVI, lymphovascular invasion; BMI, body mass index; SATI, subcutaneous adipose tissue index; VATI, visceral adipose tissue index; PFS, progression-free survival; HR, hazard ratio; CI, confidence interval.
* P < 0.05

Subsequently, a basic clinical model was developed to predict survival using stage, LVI, and BMI. Despite LVI not being an independent predictor, it was included in the clinical model due to its biological relevance in cancer progression. Additionally, the number of positive lymph nodes, associated with stage, was excluded from multivariate analysis. Significant independent predictors of prognosis from body composition, namely SMI and VATI, were incorporated into the clinical model. Each added feature enhanced the model's accuracy in predicting both PFS and OS.

A nomogram was developed by assigning a score to each clinical, body composition index, and radiomic parameter based on its prognostic impact on patient outcomes (Fig. 4–5). The C-indices for the multiparametric models predicting PFS and OS were 0.810 (95% CI: 0.737–0.884) and 0.806 (95% CI: 0.720–0.891), respectively. These values exceeded those of other models (clinical + body composition index vs. clinical alone), as shown in Table 4.

Table 4

The Harrell’s C-index results
	Progression-free survival
	C-index	95%CI	C-index	95%CI
Clinical + Body composition index + Rad-score	0.810	0.737–0.884	0.806	0.720–0.891
Clinical + Body composition index	0.782	0.670–0.864	0.752	0.660–0.844
Clinical	0.666	0.571–0.761	0.619	0.510–0.727

In this study, we assessed the prognostic value of ¹⁸F-FDG PET/CT radiomics alongside body composition indices and clinicopathological parameters in EC patients. Our findings identified Rad-score, SMI, and VATI as independent predictors of both PFS and OS. Integrating clinicopathological data with imaging-derived parameters, such as radiomics and body composition status, to develop new composite biomarkers yielded superior prognostic outcomes compared to using these factors individually.

Recent literature has seen a surge in publications exploring functional parameters from standard ¹⁸F-FDG PET/CT, including metabolic tumor volume (MTV), total lesion glycolysis (TLG), and radiomics, highlighting their significance in prognostic assessments across various cancers ^[25–29]. Radiomics, which extracts extensive high-dimensional data from medical images, offers the potential to investigate imaging heterogeneity beyond visual perception ^[30]. For instance, Amrane et al. have retrospectively studied 97 patients with gastroesophageal junction cancer (GEJC) and found that increased histogram entropy of pre-treatment PET-measured tumor correlates with poorer RFS and OS ^[31]. Similarly, another study has indicated that higher kurtosis in CT and PET components is associated with a worse prognosis in patients with esophagogastric adenocarcinoma ^[17]. These findings underscore the potential of identifying tumor heterogeneity through imaging data to enhance treatment outcomes. While there are parallels with previous studies, our research uniquely focused on patients with ESCC, contributing to a more homogeneous cohort.

In our study, potential predictors were selected using the LASSO method to construct the Rad-score, which incorporates biologically informative features, including texture features (e.g., GLCM, GLRLM, and GLZLM), conventional features (SUV_mean), and three-dimensional shape features. Zhu et al. have identified 16 features associated with distant metastasis in EC patients out of 851 radiomic features, primarily from shape, GLCM, GLSZM, GLDM, and NGTDM categories ^[32], aligning closely with our findings.

Previous research demonstrates that three-dimensional shape features (e.g., irregularity and sphericity) extracted from PET and low-dose CT images exhibit high reproducibility and accurately reflect lesion morphology ^[33]. Moreover, these high-dimensional image features enable the quantification of multi-frequency tumor information and show promise in predicting pathologic type, lymph node metastasis, and prognosis ^{[6, 34, 35]}. Our study specifically revealed that a higher Rad-score correlated with a poorer prognosis.

As previously noted, body composition parameters have been recognized as predictors in cancer patients ^[36–38]. In our study, sarcopenia predicted significantly lower PFS (HR = 5.742, 95% CI: 2.349–14.037, P < 0.001) and OS (HR = 4.477, 95% CI: 1.704–11.759, P = 0.002). Additionally, AT plays a crucial role in ESCC; patients with a high VATI had markedly shorter median PFS (HR = 6.663, 95% CI: 2.806–15.821, P < 0.001) and OS (HR = 0.323, 95% CI: 0.111–0.935, P = 0.037) compared to those with lower VATI.

The nomogram integrates multiple high-risk variables into a quantitative model, surpassing traditional tools in predicting disease onset and progression ^{[39, 40]}. However, to our knowledge, there is limited literature on predicting prognosis in ESCC using PET/CT imaging parameters and body composition indicators. Our study demonstrated that the multiparametric nomogram model, incorporating stage, LVI, BMI, SMI, VATI, and Rad-score, achieved superior C-index values compared to clinical or combined clinical-body composition index models (PFS: 0.810 vs. 0.666 vs. 0.782; OS: 0.806 vs. 0.661 vs. 0.752).

This study has several limitations. Firstly, it was a retrospective study conducted at a single institution, focusing exclusively on ESCC, with a limited sample size. Secondly, the relatively small cohort size precluded external validation of our findings. Therefore, future research collaborations with external institutions are warranted to validate our results in a larger and more diverse cohort.

Combined radiomic features extracted from baseline ¹⁸F-FDG PET/CT scans, integrating both PET and CT data, demonstrated promising prognostic potential for patients diagnosed with ESCC. Moreover, the inclusion of ¹⁸F-FDG PET/CT-derived body composition parameters alongside clinical data and radiomics enhanced prognostic predictions synergistically. This integrated approach offered a comprehensive framework to refine prognostic assessment, highlighting the transformative impact of multimodal imaging in personalized cancer care.

Acknowledgements

Not applicable.

Author contributions

DSM, ZB and YY conceived and supervised the study, ZYY, ZJ and CXW participated in data analysis and figure preparation, GSS and SSB were involved in data collection and manuscript review. All authors have read and approved the final manuscript.

Funding

This research received funding from various sources, including the National Natural Science Foundation of China (Grant Nos. 81601522 and 61927801), Gusu Health Talent Program (No. GSWS2020013), Project of Medical Research Fund of Jiangsu Provincial Health Commission (No. H2023020), Suzhou Science and Education Youth Project (No. KJXW2021004), Project of State Key Laboratory of Radiation Medicine and Protection, Soochow University (Nos. GZK1202127 and GZK12023035), NHC Key Laboratory of Nuclear Technology Medical Transformation (MIANYANG CENTRAL HOSPITAL) (Nos. 2021HYX023, 2021HYX029 and 2022HYX0009), Project of the National Engineering Research Center of Isotopes and Drugs (TWSCX2023CXJJ1101), The Natural Science Foundation of the Jiangsu Higher Education Institutions of China (No. 23KJD320001) , and The Suqian Sci & Tech Program (Grant No. K202107).

Availability of data and materials

The analyzed datasets generated during the study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the ethics committee of The First Affiliated Hospital of Soochow University (2024-298), and the requirement for informed consent was waived.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin. 2021;71(3):209-249. doi:10.3322/caac.21660
Wei WQ, Hao CQ, Guan CT, et al. Esophageal Histological Precursor Lesions and Subsequent 8.5-Year Cancer Risk in a Population-Based Prospective Study in China. Am J Gastroenterol. 2020;115(7):1036-1044. doi:10.14309/ajg. 0000000000000640
di Pietro M, Canto MI, Fitzgerald RC. Endoscopic Management of Early Adenocarcinoma and Squamous Cell Carcinoma of the Esophagus: Screening, Diagnosis, and Therapy. Gastroenterology. 2018;154(2):421-436. doi: 10.1053/j. gastro.2017.07.041
Li X, Chen L, Luan S, et al. The development and progress of nanomedicine for esophageal cancer diagnosis and treatment. Semin Cancer Biol. 2022;86(Pt 2):873-885. doi: 10.1016/j.semcancer.2022.01.007
Li L, Li W, Xu D, et al. Association Between Visceral Fat Area and Cancer Prognosis: A Population-Based Multicenter Prospective Study. Am J Clin Nutr. 2023;118(3):507-517. doi: 10.1016/j.ajcnut.2023.07.001
Hinzpeter R, Mirshahvalad SA, Kulanthaivelu R, et al. Prognostic Value of [18F]-FDG PET/CT Radiomics Combined with Sarcopenia Status among Patients with Advanced Gastroesophageal Cancer. Cancers (Basel). 2022;14(21):5314. Published 2022 Oct 28. doi:10.3390/cancers14215314
Anconina R, Ortega C, Metser U, et al. Influence of sarcopenia, clinical data, and 2-[¹⁸F] FDG PET/CT in outcome prediction of patients with early-stage adenocarcinoma esophageal cancer. Eur J Nucl Med Mol Imaging. 2022;49(3): 1012-1020. doi:10.1007/s00259-021-05514-w
Saalfeld S, Kreher R, Hille G, et al. Prognostic role of radiomics-based body composition analysis for the 1-year survival for hepatocellular carcinoma patients. J Cachexia Sarcopenia Muscle. 2023;14(5):2301-2309. doi:10.1002/jcsm. 13315
Mariette C, De Botton ML, Piessen G. Surgery in esophageal and gastric cancer patients: what is the role for nutrition support in your daily practice? Ann Surg Oncol 2012; 19:2128e34. https://doi.org/10.1245/s10434-012-2225-6.
Jordan T., Mastnak D.M., Palamar N., Kozjek N.R. Nutritional Therapy for Patients with Esophageal Cancer. Nutr. Cancer. 2018; 70:23–29. doi: 10. 1080/01635581.2017.1374417.
Yuan S, Larsson SC. Epidemiology of sarcopenia: Prevalence, risk factors, and consequences. Metabolism. 2023; 144:155533. doi: 10.1016/j.metabol.2023. 155533
Morishita, S. Prevalence of Sarcopenia in Cancer Patients: Review and Future Directions. Int. J. Phys. Med. Rehabil. 2016, 4, 1000342
Pei X, Xie Y, Liu Y, et al. Imaging-based adipose biomarkers for predicting clinical outcomes of cancer patients treated with immune checkpoint inhibitors: a systematic review. Front Oncol. 2023; 13:1198723. Published 2023 Oct 17. doi:10.3389/fonc.2023.1198723
Lee JW, Lee SM. Radiomics in Oncological PET/CT: Clinical Applications. Nucl Med Mol Imaging. 2018;52(3):170-189. doi:10.1007/s13139-017-0500-y
De Bari B, Lefevre L, Henriques J, et al. Could 18-FDG PET-CT Radiomic Features Predict the Locoregional Progression-Free Survival in Inoperable or Unresectable Oesophageal Cancer?. Cancers (Basel). 2022;14(16):4043. Published 2022 Aug 22. doi:10.3390/cancers14164043
Sah BR, Owczarczyk K, Siddique M, Cook GJR, Goh V. Radiomics in esophageal and gastric cancer. Abdom Radiol (NY). 2019;44(6):2048-2058. doi:10.1007/s00 261-018-1724-8
Anconina R, Ortega C, Metser U, et al. Combined 18 F-FDG PET/CT Radiomics and Sarcopenia Score in Predicting Relapse-Free Survival and Overall Survival in Patients With Esophagogastric Cancer. Clin Nucl Med. 2022;47(8):684-691. doi:10.1097/RLU.0000000000004253
Lu N, Zhang WJ, Dong L, et al. Dual-region radiomics signature: Integrating primary tumor and lymph node computed tomography features improves survival prediction in esophageal squamous cell cancer. Comput Methods Programs Biomed. 2021; 208:106287. doi: 10.1016/j.cmpb.2021.106287
Lucia F, Louis T, Cousin F, et al. Multicentric development and evaluation of [¹⁸F]FDG PET/CT and CT radiomic models to predict regional and/or distant recurrence in early-stage non-small cell lung cancer treated by stereotactic body radiation therapy. Eur J Nucl Med Mol Imaging. 2024;51(4):1097-1108. doi:10.1007/s00259-023-06510-y
Dissaux G, Visvikis D, Da-Ano R, et al. Pretreatment ¹⁸F-FDG PET/CT Radiomics Predict Local Recurrence in Patients Treated with Stereotactic Body Radiotherapy for Early-Stage Non-Small Cell Lung Cancer: A Multicentric Study. J Nucl Med. 2020;61(6):814-820. doi:10.2967/jnumed.119.228106
Irving BA, Weltman JY, Brock DW, Davis CK, Gaesser GA, Weltman A. NIH ImageJ and Slice-O-Matic computed tomography imaging software to quantify soft tissue. Obesity (Silver Spring). 2007;15(2):370-376. doi:10.1038/oby. 2007.573
Cespedes Feliciano EM, Popuri K, Cobzas D, et al. Evaluation of automated computed tomography segmentation to assess body composition and mortality associations in cancer patients. J Cachexia Sarcopenia Muscle. 2020;11(5):1258-1269. doi:10.1002/jcsm.12573
Derstine BA, Holcombe SA, Ross BE, Wang NC, Su GL, Wang SC. Skeletal muscle cutoff values for sarcopenia diagnosis using T10 to L5 measurements in a healthy US population. Sci Rep. 2018 Jul 27;8(1):11369. doi: 10.1038/s41598-018-29825-5.
Nioche C, Orlhac F, Boughdad S, et al. LIFEx: A Freeware for Radiomic Feature Calculation in Multimodality Imaging to Accelerate Advances in the Characterization of Tumor Heterogeneity. Cancer Res. 2018;78(16):4786-4789. doi: 10.1158/0008-5472.CAN-18-0125
Rice TW, Patil DT, Blackstone EH. 8th edition AJCC/UICC staging of cancers of the esophagus and esophagogastric junction: application to clinical practice. Ann Cardiothorac Surg. 2017;6(2):119-130. doi:10.21037/acs.2017.03.14
Hatt M, Tixier F, Pierce L, Kinahan PE, Le Rest CC, Visvikis D. Characterization of PET/CT images using texture analysis: the past, the present… any future?. Eur J Nucl Med Mol Imaging. 2017;44(1):151-165. doi:10.1007/s00259-016-3427-0
Carlier T, Frécon G, Mateus D, et al. Prognostic Value of ¹⁸F-FDG PET Radiomics Features at Baseline in PET-Guided Consolidation Strategy in Diffuse Large B-Cell Lymphoma: A Machine-Learning Analysis from the GAINED Study. J Nucl Med. 2024;65(1):156-162. Published 2024 Jan 2. doi:10.2967/jnumed.123.265872
Wang P, Luo Z, Luo C, Wang T. Application of a Comprehensive Model Based on CT Radiomics and Clinical Features for Postoperative Recurrence Risk Prediction in Non-small Cell Lung Cancer. Acad Radiol. Published online January 2, 2024. doi: 10.1016/j.acra.2023.11.028
Huang W, Li L, Liu S, et al. Enhanced CT-based radiomics predicts pathological complete response after neoadjuvant chemotherapy for advanced adenocarcinoma of the esophagogastric junction: a two-center study. Insights Imaging. 2022;13(1):134. Published 2022 Aug 17. doi:10.1186/s13244-022-01273-w
Ak M, Toll SA, Hein KZ, Colen RR, Khatua S. Evolving Role and Translation of Radiomics and Radiogenomics in Adult and Pediatric Neuro-Oncology. AJNR Am J Neuroradiol. 2022;43(6):792-801. doi:10.3174/ajnr. A7297
Amrane K, Thuillier P, Bourhis D, et al. Prognostic value of pre-therapeutic FDG-PET radiomic analysis in gastro-esophageal junction cancer [published correction appears in Sci Rep. 2023 Apr 24;13(1):6665]. Sci Rep. 2023;13(1):5789. Published 2023 Apr 8. doi:10.1038/s41598-023-31587-8
Zhu C, Mu F, Wang S, Qiu Q, Wang S, Wang L. Prediction of distant metastasis in esophageal cancer using a radiomics-clinical model. Eur J Med Res. 2022 Dec 3;27(1):272. doi: 10.1186/s40001-022-00877-8. PMID: 36463269; PMCID: PMC9719117.
Desseroit MC, Tixier F, Weber WA, et al. Reliability of PET/CT Shape and Heterogeneity Features in Functional and Morphologic Components of Non-Small Cell Lung Cancer Tumors: A Repeatability Analysis in a Prospective Multicenter Cohort. J Nucl Med. 2017;58(3):406-411. doi:10.2967/jnumed. 116.180919
Cui J, Li L, Liu N, Hou W, Dong Y, Yang F, Zhu S, Li J, Yuan S. Model integrating CT-based radiomics and genomics for survival prediction in esophageal cancer patients receiving definitive chemoradiotherapy. Biomark Res. 2023 Apr 24;11(1):44. doi: 10.1186/s40364-023-00480-x. PMID: 37095586; PMCID: PMC10127317.
Lei X, Cao Z, Wu Y, Lin J, Zhang Z, Jin J, Ai Y, Zhang J, Du D, Tian Z, Xie C, Yin W, Jin X. Preoperative prediction of clinical and pathological stages for patients with esophageal cancer using PET/CT radiomics. Insights Imaging. 2023 Oct 15;14(1):174. doi: 10.1186/s13244-023-01528-0. PMID: 37840068; PMCID: PMC10577114.
Iyer K, Beeche CA, Gezer NS, et al. CT-Derived Body Composition Is a Predictor of Survival after Esophagectomy. J Clin Med. 2023;12(6):2106. Published 2023 Mar 8. doi:10.3390/jcm12062106
Fu Y, Wang X, Yi X, et al. Ensemble Machine Learning Model Incorporating Radiomics and Body Composition for Predicting Intraoperative HDI in PPGL. J Clin Endocrinol Metab. 2024;109(2):351-360. doi:10.1210/clinem/dgad543
Shi H, Wei Y, Cheng S, et al. Survival prediction after upfront surgery in patients with pancreatic ductal adenocarcinoma: Radiomic, clinic-pathologic and body composition analysis. Pancreatology. 2021;21(4):731-737. doi: 10.1016/j.pan. 2021.02.009
Dong D, Fang MJ, Tang L, et al. Deep learning radiomic nomogram can predict the number of lymph node metastasis in locally advanced gastric cancer: an international multicenter study. Ann Oncol. 2020;31(7):912-920. doi: 10.1016/j.annonc.2020.04.003
Gu B, Meng M, Xu M, et al. Multi-task deep learning-based radiomic nomogram for prognostic prediction in locoregionally advanced nasopharyngeal carcinoma. Eur J Nucl Med Mol Imaging. 2023;50(13):3996-4009. doi:10.1007/s00259-023-06399-7

No competing interests reported.

Download PDF

Reviewers invited by journal
31 Jul, 2024
Editor invited by journal
31 Jul, 2024
Editor assigned by journal
31 Jul, 2024
Submission checks completed at journal
31 Jul, 2024
First submitted to journal
30 Jul, 2024

You are reading this latest preprint version

Integrating 18 F-FDG PET/CT Radiomics and Body Composition for Enhanced Prognostic Assessment in Patients with Esophageal Cancer

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials and methods

2.1. Patient data

2.2. Image acquisition

2.3. Body composition measurements

2.4. Tumor segmentation and feature extraction

2.5. Feature selection and Rad-score construction

2.6. Statistical analysis

3. Results

3.1. Patient characteristics

3.2. Factors associated with body composition index

3.3. Prognostic impact

Discussion

Conclusions

Declarations

References

Additional Declarations

Status:

Version 1