Deep neural network for automatic volumetric segmentation of whole-body CT images for body composition assessment

Body Composition to Define Prognosis of Cancers Treated by Anti-Angiogenic Drugs

Background: Body composition could help to better define the prognosis of cancers treated with anti-angiogenics. The aim of this study is to evaluate the prognostic value of 3D and 2D anthropometric parameters in patients given anti-angiogenic treatments. Methods: 526 patients with different types of cancers were retrospectively included. The software Anthropometer3DNet was used to measure automatically fat body mass (FBM3D), muscle body mass (MBM3D), visceral fat mass (VFM3D) and subcutaneous fat mass (SFM3D) in 3D computed tomography. For comparison, equivalent two-dimensional measurements at the L3 level were also measured. The area under the curve (AUC) of the receiver operator characteristics (ROC) was used to determine the parameters’ predictive power and optimal cut-offs. A univariate analysis was performed using Kaplan−Meier on the overall survival (OS). Results: In ROC analysis, all 3D parameters appeared statistically significant: VFM3D (AUC = 0.554, p = 0.02, cutoff = 0.72 kg/m2), SFM3D (AUC = 0.544, p = 0.047, cutoff = 3.05 kg/m2), FBM3D (AUC = 0.550, p = 0.03, cutoff = 4.32 kg/m2) and MBM3D (AUC = 0.565, p = 0.007, cutoff = 5.47 kg/m2), but only one 2D parameter (visceral fat area VFA2D AUC = 0.548, p = 0.034). In log-rank tests, low VFM3D (p = 0.014), low SFM3D (p < 0.0001), low FBM3D (p = 0.00019) and low VFA2D (p = 0.0063) were found as a significant risk factor. Conclusion: automatic and 3D body composition on pre-therapeutic CT is feasible and can improve prognostication in patients treated with anti-angiogenic drugs. Moreover, the 3D measurements appear to be more effective than their 2D counterparts.

Automated segmentation of five different body tissues on computed tomography using deep learning

PURPOSE

To develop and validate a computer tool for automatic and simultaneous segmentation of five body tissues depicted on computed tomography (CT) scans: visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT), intermuscular adipose tissue (IMAT), skeletal muscle (SM), and bone.

METHODS

A cohort of 100 CT scans acquired on different subjects were collected from The Cancer Imaging Archive-50 whole-body positron emission tomography-CTs, 25 chest, and 25 abdominal. Five different body tissues (i.e., VAT, SAT, IMAT, SM, and bone) were manually annotated. A training-while-annotating strategy was used to improve the annotation efficiency. The 10-fold cross-validation method was used to develop and validate the performance of several convolutional neural networks (CNNs), including UNet, Recurrent Residual UNet (R2Unet), and UNet++. A grid-based three-dimensional patch sampling operation was used to train the CNN models. The CNN models were also trained and tested separately for each body tissue to see if they could achieve a better performance than segmenting them jointly. The paired sample t-test was used to statistically assess the performance differences among the involved CNN models RESULTS: When segmenting the five body tissues simultaneously, the Dice coefficients ranged from 0.826 to 0.840 for VAT, from 0.901 to 0.908 for SAT, from 0.574 to 0.611 for IMAT, from 0.874 to 0.889 for SM, and from 0.870 to 0.884 for bone, which were significantly higher than the Dice coefficients when segmenting the body tissues separately (p < 0.05), namely, from 0.744 to 0.819 for VAT, from 0.856 to 0.896 for SAT, from 0.433 to 0.590 for IMAT, from 0.838 to 0.871 for SM, and from 0.803 to 0.870 for bone.

CONCLUSION

There were no significant differences among the CNN models in segmenting body tissues, but jointly segmenting body tissues achieved a better performance than segmenting them separately.

Low skull bone density is associated with poor motor prognosis in women with Parkinson’s disease

Parkinson’s disease (PD) and osteoporosis are degenerative diseases that have shared pathomechanisms. To investigate the associations of skull bone density with nigrostriatal dopaminergic degeneration and longitudinal motor prognosis in female patients with PD. We analyzed the data of 260 drug-naïve female PD patients aged ≥50 years old who were followed-up for ≥3 years after their first visit to the clinic with baseline dopamine transporter (DAT) imaging. We measured skull bone density as a surrogate marker for systemic bone loss by calculating the Hounsfield unit (HU) in computed tomography scans. A Cox proportional hazard model was built to compare the rates of levodopa-induced dyskinesia (LID) or wearing-off according to skull HU. Longitudinal changes in levodopa-equivalent dose (LED) during a 3-year follow-up were assessed using a linear mixed model. A lower skull HU was associated with lower baseline DAT availability in striatal subregions; however, this relationship was not significant after adjusting for age, disease duration, body mass index, and white matter hyperintensities. After adjusting for confounding factors, a lower skull HU was significantly associated with an increased risk of LID development (hazard ratio = 1.660 per 1 standard deviation decrease, p = 0.007) and wearing-off (hazard ratio = 1.613, p = 0.016) in younger (&lt;67 years) but not in older patients. Furthermore, a lower skull HU was associated with a steeper increase in LED during follow-up in younger patients only (β = –21.99, p &lt; 0.001). This study suggests that baseline skull bone density would be closely linked to motor prognosis in drug naïve women with PD.

Combining Deep Learning and Radiomics for Automated, Objective, Comprehensive Bone Marrow Characterization From Whole-Body MRI: A Multicentric Feasibility Study

OBJECTIVES

Disseminated bone marrow (BM) involvement is frequent in multiple myeloma (MM). Whole-body magnetic resonance imaging (wb-MRI) enables to evaluate the whole BM. Reading of such whole-body scans is time-consuming, and yet radiologists can transfer only a small fraction of the information of the imaging data set to the report. This limits the influence that imaging can have on clinical decision-making and in research toward precision oncology. The objective of this feasibility study was to implement a concept for automatic, comprehensive characterization of the BM from wb-MRI, by automatic BM segmentation and subsequent radiomics analysis of 30 different BM spaces (BMS).

MATERIALS AND METHODS

This retrospective multicentric pilot study used a total of 106 wb-MRI from 102 patients with (smoldering) MM from 8 centers. Fifty wb-MRI from center 1 were used for training of segmentation algorithms (nnU-Nets) and radiomics algorithms. Fifty-six wb-MRI from 8 centers, acquired with a variety of different MRI scanners and protocols, were used for independent testing. Manual segmentations of 2700 BMS from 90 wb-MRI were performed for training and testing of the segmentation algorithms. For each BMS, 296 radiomics features were calculated individually. Dice score was used to assess similarity between automatic segmentations and manual reference segmentations.

RESULTS

The "multilabel nnU-Net" segmentation algorithm, which performs segmentation of 30 BMS and labels them individually, reached mean dice scores of 0.88 ± 0.06/0.87 ± 0.06/0.83 ± 0.11 in independent test sets from center 1/center 2/center 3-8 (interrater variability between radiologists, 0.88 ± 0.01). The subset from the multicenter, multivendor test set (center 3-8) that was of high imaging quality was segmented with high precision (mean dice score, 0.87), comparable to the internal test data from center 1. The radiomic BM phenotype consisting of 8880 descriptive parameters per patient, which result from calculation of 296 radiomics features for each of the 30 BMS, was calculated for all patients. Exemplary cases demonstrated connections between typical BM patterns in MM and radiomic signatures of the respective BMS. In plausibility tests, predicted size and weight based on radiomics models of the radiomic BM phenotype significantly correlated with patients' actual size and weight ( P = 0.002 and P = 0.003, respectively).

CONCLUSIONS

This pilot study demonstrates the feasibility of automatic, objective, comprehensive BM characterization from wb-MRI in multicentric data sets. This concept allows the extraction of high-dimensional phenotypes to capture the complexity of disseminated BM disorders from imaging. Further studies need to assess the clinical potential of this method for automatic staging, therapy response assessment, or prediction of biopsy results.

Remotely shared CT-derived presurgical understanding of lung cancer: A randomized trial

Shared decision‐making is imperative for patient‐and family‐centered care. However, gathering individuals in a single place was challenged by modern life and pandemic restrictions. This study conducted a 1:1 randomized trial to examine the feasibility of a CT‐derived 3D virtual explanation module for lung cancer to improve the understanding of patients and third parties in physically separate locations. We prospectively enrolled adults in whom elective surgical resection for lung cancer was planned at a single tertiary hospital in 2020. From presurgical CT scans, deep neural networks automatically segmented lung cancer, airway, pulmonary lobes, skin, and bony thorax. The segmented structures were subsequently transformed into an anonymized interactive 3D module which comprised a standardized scenario with explanatory texts. The intervention group received a link to the module on their smartphone before admission and could repeatedly access the link or transfer it to patients' third parties. A total of 33 and 29 patients were enrolled in the intervention and control arms. The understanding score did not statistically differ between the arms (mean difference, 0.7 [95% CI: −0.2, 1.5]; p = 0.13). However, 76% of patients in the intervention arm accessed the link, and patient median access count was 14. The link recipients of third parties had comparable understanding scores to the patients (mean difference, −0.2 [95% CI: −1.9, 1.5]; p = 1.00), indicating that the understanding could be shared remotely with patients and patients’ third parties. In conclusion, it was feasible that people physically separated from patients obtained a comparable understanding of lung cancer surgery using the patient's CT‐derived 3D virtual explanation module.

Preoperative Thoracic Muscle Mass Predicts Bone Density Change After Parathyroidectomy in Primary Hyperparathyroidism

CONTEXT

Predicting bone mineral density (BMD) gain after parathyroidectomy may influence individualized therapeutic approaches for treating patients with primary hyperparathyroidism (PHPT).

OBJECTIVE

This study aimed to assess whether skeletal muscle mass data could predict BMD change after parathyroidectomy in patients with PHPT.

METHODS

This retrospective study collected data from 2012 to 2021 at Severance Hospital, Seoul, Korea. A total of 130 patients (mean age, 64.7 years; 81.5% women) with PHPT who underwent parathyroidectomy were analyzed. Thoracic muscle volume (T6-T7 level) was estimated using noncontrast parathyroid single photon emission computed tomography/computed tomography (SPECT/CT) scans and an automated deep-learning-based software. The primary outcome assessed was the change in femoral neck BMD (FNBMD, %) 1 year after parathyroidectomy.

RESULTS

The median degree of FNBMD change after parathyroidectomy was + 2.7% (interquartile range: -0.9 to + 7.6%). Elevated preoperative PTH level was associated with lower thoracic muscle mass (adjusted β: -8.51 cm3 per one log-unit PTH increment, P = .045) after adjusting for age, sex, body mass index (BMI), and baseline FNBMD. One SD decrement in thoracic muscle mass was associated with lesser FNBMD (adjusted β: -2.35%, P = .034) gain and lumbar spine BMD gain (adjusted β: -2.51%, P = .044) post surgery after adjusting for covariates.

CONCLUSION

Lower thoracic skeletal muscle mass was associated with elevated preoperative PTH levels in patients with PHPT. Lower skeletal muscle mass was associated with lesser BMD gain after parathyroidectomy, independent of age, sex, BMI, preoperative BMD, and PTH level.

Measurement of Sarcopenia in Head and Neck Cancer Patients and Its Association With Frailty

In head and neck cancer (HNC) there is a need for more personalized treatment based on risk assessment for treatment related adverse events (i.e. toxicities and complications), expected survival and quality of life. Sarcopenia, defined as a condition characterized by loss of skeletal muscle mass and function, can predict adverse outcomes in HNC patients. A review of the literature on the measurement of sarcopenia in head and neck cancer patients and its association with frailty was performed. Skeletal muscle mass (SMM) measurement only is often used to determine if sarcopenia is present or not. SMM is most often assessed by measuring skeletal muscle cross-sectional area on CT or MRI at the level of the third lumbar vertebra. As abdominal scans are not always available in HNC patients, measurement of SMM at the third cervical vertebra has been developed and is frequently used. Frailty is often defined as an age-related cumulative decline across multiple physiologic systems, with impaired homeostatic reserve and a reduced capacity of the organism to withstand stress, leading to increased risk of adverse health outcomes. There is no international standard measure of frailty and there are multiple measures of frailty. Both sarcopenia and frailty can predict adverse outcomes and can be used to identify vulnerable patients, select treatment options, adjust treatments, improve patient counselling, improve preoperative nutritional status and anticipate early on complications, length of hospital stay and discharge. Depending on the definitions used for sarcopenia and frailty, there is more or less overlap between both conditions. However, it has yet to be determined if sarcopenia and frailty can be used interchangeably or that they have additional value and should be used in combination to optimize individualized treatment in HNC patients.

Automated segmentation of whole-body CT images for body composition analysis in pediatric patients using a deep neural network

OBJECTIVES

To develop an automatic segmentation algorithm using a deep neural network with transfer learning applicable to whole-body PET-CT images in children.

METHODS

For model development, we utilized transfer learning with a pre-trained model based on adult patients. We used CT images of 31 pediatric patients under 19 years of age (mean age, 9.6 years) who underwent PET-CT from institution #1 for transfer learning. Two radiologists manually labeled the skin, bone, muscle, abdominal visceral fat, subcutaneous fat, internal organs, and central nervous system in each CT slice and used these as references. For external validation, we collected 14 pediatric PET/CT scans from institution #2 (mean age, 9.1 years). The Dice similarity coefficients (DSCs), sensitivities, and precision were compared between the algorithms before and after transfer learning. In addition, we evaluated segmentation performance according to sex, age (≤ 8 vs. > 8 years), and body mass index (BMI, ≤ 20 vs. > 20 kg/m²).

RESULTS

The algorithm after transfer learning showed better performance than the algorithm before transfer learning for all body compositions (p < 0.001). The average DSC, sensitivity, and precision of each algorithm before and after transfer learning were 98.23% and 99.28%, 98.16% and 99.28%, and 98.29% and 99.28%, respectively. The segmentation performance of the algorithm was generally not affected by age, sex, or BMI, except for precision in the body muscle compartment.

CONCLUSION

The developed model with transfer learning enabled accurate and fully automated segmentation of multiple tissues on whole-body CT scans in children.

KEY POINTS

• We utilized transfer learning with a pre-trained segmentation algorithm for adult to develop an algorithm for automated segmentation of pediatric whole-body CT. • This algorithm showed excellent performance and was not affected by sex, age, or body mass index, except for precision in body muscle.

Adrenal Morphology as an Indicator of Long-Term Disease Control in Adults with Classic 21-Hydroxylase Deficiency

BACKGROUND

Monitoring adults with classical 21-hydroxylase deficiency (21OHD) is challenging due to variation in clinical and laboratory settings. Moreover, guidelines for adrenal imaging in 21OHD are not yet available. We evaluated the relationship between adrenal morphology and disease control status in classical 21OHD.

METHODS

This retrospective, cross-sectional study included 90 adult 21OHD patients and 270 age- and sex-matched healthy controls. We assessed adrenal volume, width, and tumor presence using abdominal computed tomography and evaluated correlations of adrenal volume and width with hormonal status. We investigated the diagnostic performance of adrenal volume and width for identifying well-controlled status in 21OHD patients (17α-hydroxyprogesterone [17-OHP] <10 ng/mL).

RESULTS

The adrenal morphology of 21OHD patients showed hypertrophy (45.6%), normal size (42.2%), and hypotrophy (12.2%). Adrenal tumors were detected in 12 patients (13.3%). The adrenal volume and width of 21OHD patients were significantly larger than those of controls (18.2±12.2 mL vs. 7.1±2.0 mL, 4.7±1.9 mm vs. 3.3±0.5 mm, P<0.001 for both). The 17-OHP and androstenedione levels were highest in patients with adrenal hypertrophy, followed by those with normal adrenal glands and adrenal hypotrophy (P<0.05 for both). Adrenal volume and width correlated positively with adrenocorticotropic hormone, 17-OHP, 11β-hydroxytestosterone, progesterone sulfate, and dehydroepiandrosterone sulfate in both sexes (r=0.33-0.95, P<0.05 for all). For identifying well-controlled patients, the optimal cut-off values of adrenal volume and width were 10.7 mL and 4 mm, respectively (area under the curve, 0.82-0.88; P<0.001 for both).

CONCLUSION

Adrenal volume and width may be reliable quantitative parameters for monitoring patients with classical 21OHD.

Prognostic role of computed tomography-based, artificial intelligence-driven waist skeletal muscle volume in uterine endometrial carcinoma

OBJECTIVES

To investigate the impact of computed tomography (CT)-based, artificial intelligence-driven waist skeletal muscle volume on survival outcomes in patients with endometrial cancer.

METHODS

We retrospectively identified endometrial cancer patients who received primary surgical treatment between 2014 and 2018 and whose pre-treatment CT scans were available (n = 385). Using an artificial intelligence-based tool, the skeletal muscle area (cm2) at the third lumbar vertebra (L3) and the skeletal muscle volume (cm3) at the waist level were measured. These values were converted to the L3 skeletal muscle index (SMI) and volumetric SMI by normalisation with body height. The relationships between L3, volumetric SMIs, and survival outcomes were evaluated.

RESULTS

Setting 39.0 cm2/m2 of L3 SMI as cut-off value for sarcopenia, sarcopenia (< 39.0 cm2/m2, n = 177) and non-sarcopenia (≥ 39.0 cm2/m2, n = 208) groups showed similar progression-free survival (PFS; p = 0.335) and overall survival (OS; p = 0.241). Using the median value, the low-volumetric SMI group (< 206.0 cm3/m3, n = 192) showed significantly worse PFS (3-year survival rate, 77.3% vs. 88.8%; p = 0.004) and OS (3-year survival rate, 92.8% vs. 99.4%; p = 0.003) than the high-volumetric SMI group (≥ 206.0 cm3/m3, n = 193). In multivariate analyses adjusted for baseline body mass index and other factors, low-volumetric SMI was identified as an independent poor prognostic factor for PFS (adjusted HR, 1.762; 95% CI, 1.051-2.953; p = 0.032) and OS (adjusted HR, 5.964; 95% CI, 1.296-27.448; p = 0.022).

CONCLUSIONS

Waist skeletal muscle volume might be a novel prognostic biomarker in patients with endometrial cancer. Assessing body composition before treatment can provide important prognostic information for such patients.

[Potential of radiomics and artificial intelligence in myeloma imaging : Development of automatic, comprehensive, objective skeletal analyses from whole-body imaging data]

CLINICAL/METHODICAL ISSUE

Multiple myeloma can affect the complete skeleton, which makes whole-body imaging necessary. With the current assessment of these complex datasets by radiologists, only a small part of the accessible information is assessed and reported.

STANDARD RADIOLOGICAL METHODS

Depending on the question and availability, computed tomography (CT), magnetic resonance imaging (MRI), or positron emission tomography (PET) is performed and the results are then visually examined by radiologists.

METHODOLOGICAL INNOVATIONS

A combination of automatic skeletal segmentation using artificial intelligence and subsequent radiomics analysis of each individual bone have the potential to provide automatic, comprehensive, and objective skeletal analyses.

PERFORMANCE

A few automatic skeletal segmentation algorithms for CT already show promising results. In addition, first studies indicate correlations between radiomics features of bone and bone marrow with established disease markers and therapy response.

ACHIEVEMENTS

Artificial intelligence (AI) and radiomics algorithms for automatic skeletal analysis from whole-body imaging are currently in an early phase of development.

Computed Tomography-Derived Skeletal Muscle Radiodensity Is an Early, Sensitive Marker of Age-Related Musculoskeletal Changes in Healthy Adults

BACKGROUND

A decrease in computed tomography (CT)-derived skeletal muscle radiodensity (SMD) reflects age-related ectopic fat infiltration of muscle, compromising muscle function and metabolism. We investigated the age-related trajectory of SMD and its association with vertebral trabecular bone density in healthy adults.

METHODS

In a cohort of healthy adult kidney donors aged 19 to 69 years (n=583), skeletal muscle index (SMI, skeletal muscle area/height2), SMD, and visceral-to-subcutaneous fat (V/S) ratio were analyzed at the level of L3 from preoperative CT scans. Low bone mass was defined as an L1 trabecular Hounsfield unit (HU) <160 HU.

RESULTS

L3SMD showed constant decline from the second decade (annual change -0.38% and -0.43% in men and women), whereas the decline of L3SMI became evident only after the fourth decade of life (-0.37% and -0.18% in men and women). One HU decline in L3SMD was associated with elevated odds of low bone mass (adjusted odds ratio, 1.07; 95% confidence interval, 1.02 to 1.13; P=0.003), independent of L3SMI, age, sex, and V/S ratio, with better discriminatory ability compared to L3SMI (area under the receiver-operating characteristics curve 0.68 vs. 0.53, P<0.001). L3SMD improved the identification of low bone mass when added to age, sex, V/S ratio, and L3SMI (category-free net reclassification improvement 0.349, P<0.001; integrated discrimination improvement 0.015, P=0.0165).

CONCLUSION

L3SMD can be an early marker for age-related musculoskeletal changes showing linear decline throughout life from the second decade in healthy adults, with potential diagnostic value for individuals with low bone mass.

Pectoralis major muscle quantification by cardiac MRI is a strong predictor of major events in HF

Clinical overt cardiac cachexia is a late ominous sign in patients with heart failure (HF) and reduced left ventricular ejection fraction (LVEF). The main goal of this study was to assess the feasibility and prognostic significance of muscle mass quantification by cardiac magnetic resonance (CMR) in HF with reduced LVEF. HF patients with LVEF < 40% (HFrEF) referred for CMR were retrospectively identified in a single center. Key exclusion criteria were primary muscle disease, known infiltrative myocardial disease and intracardiac devices. Pectoralis major muscles were measured on standard axial images at the level of the 3rd rib anteriorly. Time to all-cause death or HF hospitalization was the primary endpoint. A total of 298 HF patients were included (mean age 64 ± 12 years; 76% male; mean LVEF 30 ± 8%). During a median follow-up of 22 months (IQR: 12-33), 67 (22.5%) patients met the primary endpoint (33 died and 45 had at least 1 HF hospitalization). In multivariate analysis, LVEF [Hazard Ratio (HR): 0.950; 95% Confidence Interval (CI): 0.917-0.983; p = 0.003), NYHA class I-II vs III-IV (HR: 0.480; CI: 0.272-0.842; p = 0.010), creatinine (HR: 2.653; CI: 1.548-4.545; p < 0.001) and pectoralis major area (HR: 0.873; 95% CI: 0.821-0.929; p < 0.001) were independent predictors of the primary endpoint, when adjusted for gender and NT-pro-BNP levels. Pectoralis major size measured by CMR in HFrEF was independently associated with a higher risk of death or HF hospitalization. Further studies to establish appropriate age and gender-adjusted cut-offs of muscle areas are needed to identify high-risk subgroups.

Impacts of body composition parameters and liver cirrhosis on the severity of alcoholic acute pancreatitis

AIM

Liver cirrhosis and features of muscle or adipose tissues may affect the severity of acute pancreatitis (AP). We aimed to evaluate the impact of body composition parameters and liver cirrhosis on the severity of AP in patients with alcohol-induced AP (AAP).

METHODS

Patients with presumed AAP who underwent CT within one week after admission were retrospectively enrolled. L3 sectional areas of abdominal fat and muscle, and mean muscle attenuations (MMAs) were quantified. The presence of liver cirrhosis was determined using clinical and CT findings. Factors potentially associated with moderately severe or severe AP were included in the multivariable logistic regression analysis.

RESULTS

A total of 242 patients (47.0 ± 12.6 years, 215 males) with presumed AAP were included. The mild and moderately severe/severe (MSS) groups included 137 (56.6%) and 105 patients (43.4%), respectively. Patients in the MSS group had higher rates of liver cirrhosis, organ failure, and local complications. Among body composition parameters, mean MMA (33.4 vs 36.8 HU, P<0.0001) and abdominal muscle mass (126.5 vs 135.1 cm2, P = 0.029) were significantly lower in the MSS group. The presence of liver cirrhosis (OR, 4.192; 95% CI, 1.620-10.848) was found to be a significant risk factor for moderately severe or severe AP by multivariable analysis.

CONCLUSION

The results of this study suggest that liver cirrhosis has a significant impact on the severity of AAP. Of the body composition parameters examined, MMA and abdominal muscle mass showed potential as promising predictors.

Association of Adipopenia at Preoperative PET/CT with Mortality in Stage I Non-Small Cell Lung Cancer

Background Body mass index (BMI) and sarcopenia status are well-established prognostic factors in patients with lung cancer. However, the relationship between the amount of adipose tissue and survival remains unclear. Purpose To investigate the association between baseline adipopenia and outcomes in patients with early-stage non-small cell lung cancer (NSCLC). Materials and Methods Consecutive patients who underwent surgical resection for stage I NSCLC between 2011 and 2015 at a single tertiary care center were retrospectively identified. The primary outcome was the 5-year overall survival (OS) rate, and secondary outcomes were the 5-year disease-free survival (DFS) rate and the major postoperative complication rate. The abdominal total fat volume at the waist and the skeletal muscle area at the L3 level were obtained from preoperative PET/CT data and were normalized by the height squared to calculate the fat volume index (FVI) and skeletal muscle index. Adipopenia was defined as the sex-specific lowest quartile of the FVI for the study sample, and sarcopenia was determined using the skeletal muscle index reference value (men, <55 cm²/m²; women, <39 cm²/m²). The association between body composition and outcomes was evaluated using Cox regression analysis. Results A total of 440 patients (median age, 65 years [interquartile range, 58-72 years]; 243 men) were evaluated. Most underweight patients (<20 kg/m²) had adipopenia (97%, 36 of 37 patients), but overweight patients (25-30 kg/m², n = 138) and obese patients (>30 kg/m², n = 14) did not have adipopenia (3%, four of 152 patients). In the group with a normal BMI (20-25 kg/m²), 28% (70 of 251 patients) had adipopenia and 67% (168 of 251 patients) had sarcopenia. After adjusting for age, sex, smoking history, surgical procedure, stage, histologic type, BMI, and sarcopenia, adipopenia was associated with reduced 5-year OS (hazard ratio [HR] = 2.2; 95% CI: 1.1, 3.8; P = .02) and 5-year non-cancer-specific OS rates (HR = 3.2; 95% CI: 1.2, 8.7; P = .02). There was no association between adipopenia and postoperative complications (P = .45) or between adipopenia and the 5-year DFS rate (P = .18). Conclusion Baseline adipopenia was associated with a reduced 5-year overall survival rate in patients with early-stage non-small cell lung cancer and may indicate risk for non-cancer-related death. © RSNA, 2021 Online supplemental material is available for this article.

Impact of Computed Tomography-Based, Artificial Intelligence-Driven Volumetric Sarcopenia on Survival Outcomes in Early Cervical Cancer

The purpose of this study was to investigate the impact of sarcopenia and body composition change during primary treatment on survival outcomes in patients with early cervical cancer. We retrospectively identified patients diagnosed with 2009 International Federation of Gynecology and Obstetrics stage IB1-IIA2 cervical cancer who underwent primary radical hysterectomy between 2007 and 2019. From pre-treatment CT scans (n = 306), the skeletal muscle area at the third lumbar vertebra (L3) and the waist skeletal muscle volume were measured using an artificial intelligence-based tool. These values were converted to the L3 and volumetric skeletal muscle indices by normalization. We defined L3 and volumetric sarcopenia using 39.0 cm2/m2 and the first quartile (Q1) value, respectively. From pre- and post-treatment CT scan images (n = 192), changes (%) in waist skeletal muscle and fat volumes were assessed. With the use of Cox regression models, factors associated with progression-free survival (PFS) and overall survival (OS) were analyzed. Between the L3 sarcopenia and non-sarcopenia groups, no differences in PFS and OS were observed. In contrast, volumetric sarcopenia was identified as a poor prognostic factor for PFS (adjusted hazard ratio [aHR], 1.874; 95% confidence interval [CI], 1.028-3.416; p = 0.040) and OS (aHR, 3.001; 95% CI, 1.016-8.869; p = 0.047). During primary treatment, significant decreases in waist skeletal muscle (median, -3.9%; p < 0.001) and total fat (median, -5.3%; p < 0.001) were observed. Of the two components, multivariate analysis revealed that the waist fat gain was associated with worse PFS (aHR, 2.007; 95% CI, 1.009-3.993; p = 0.047). The coexistence of baseline volumetric sarcopenia and waist fat gain further deteriorated PFS (aHR, 2.853; 95% CI, 1.257-6.474; p = 0.012). In conclusion, baseline volumetric sarcopenia might be associated with poor survival outcomes in patients with early cervical cancer undergoing primary RH. Furthermore, sarcopenia patients who gained waist fat during primary treatment were at a high risk of disease recurrence.