Body composition analysis using CT at three aspects of the lumbar third vertebra and its impact on the diagnosis of sarcopenia.

Introduction
Sarcopenia is a syndrome characterized by a progressive decrease in skeletal muscle mass, strength, and physical performance with aging [1]. Both the European Working Group on Sarcopenia in Older People (EWGSOP) and the Asian Working Group for Sarcopenia (AWGS) have clarified the diagnosis of skeletal muscle strength, including convenient measurement methods and corresponding cutoff values [2, 3]. However, the diagnostic method for skeletal muscle mass has certain limitations, as its measurement methods, such as dual-energy X-ray absorptiometry (DXA) or bioelectrical impedance analysis (BIA), are not easily available in clinical practice. Additionally, the accuracy of DXA and BIA measurements depends largely on the accuracy of the equipment equations and evaluation conditions, such as temperature, humidity, and skin conditions. Therefore, there is a need for more convenient, accessible, and accurate methods for diagnosing muscle mass. Since the gold standard for skeletal muscle mass is derived from whole-body computed tomography (CT) [4], in clinical practice, the whole body of a patient is rarely scanned. Thus, we generally adopt the aspect of muscle area on CT to reflect whole-body muscle mass [5, 6]. The skeletal muscle area (SMA) of the thoracic 12th vertebra CT imaging and lumbar 3rd vertebra (L3) CT imaging have been widely applied to evaluate muscle mass in sarcopenia patients, and their correlation with whole-body muscle mass has been confirmed by several studies [7–11].
Gastric cancer is the fifth most common malignancy and the fourth leading cause of cancer-related death worldwide and is mainly treated by curative gastrectomy [12]. Several studies have found that sarcopenia is an independent risk factor for postoperative complications and overall survival in patients with gastric cancer [13–17]. Patients with gastric cancer routinely undergo preoperative CT of the abdomen to assess gastric cancer staging. Thus, the SMA can be measured via L3-CT imaging without additional examinations to facilitate the clinical assessment of sarcopenia.
Several research teams have used different L3-CT imaging cross-sections to diagnose skeletal muscle mass. Carey et al. used the superior aspect [18], Martin et al. intercepted the transverse aspect [19], and Zhuang et al. used the inferior aspect as the foundation for their diagnoses [14]. Moreover, certain teams lack precise representations of the chosen cross-sections [20], while others have intercepted aspects that utilize the cutoffs of other teams that are inconsistent with the aspect from which the cutoff originates [15, 21, 22]. This study compared data, including muscle area, on three aspects of L3-CT imaging and compared prognostic differences in the diagnosis of sarcopenia for the first time using different L3-CT imaging aspects.

Materials and methods

Patients
This study was approved by the Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University. All patients who underwent radical gastrectomy for gastric cancer in the Department of Gastroenterology, The First Affiliated Hospital of Wenzhou Medical University, between July 2014 and February 2019 were included in this study. The inclusion criteria were as follows: (1) planned to undergo gastrectomy with curative intent for gastric cancer; (2) had preoperative abdominal CT available for review (no more than 1 month prior to surgery); and (3) agreed to participate in this study and signed an informed consent form. The exclusion criteria included (1) having physical deformities that prevented them from performing muscle strength or physical fitness tests and (2) those who underwent palliative surgery. All patients were routinely managed according to the 2010 Japanese Gastric Cancer Treatment Guidelines (ver. 3) [23]. All procedures were performed by experienced surgeons who independently performed > 200 standard gastric cancer surgeries.

Follow-up
All patients were followed up within 1 month of surgery. Thereafter, patients were followed-up every 3 months for 2 years, every 6 months thereafter for up to 5 years, and every 1 year thereafter. Patients were contacted by telephone and scheduled to return to the hospital at the aforementioned time points to complete the follow-up program. The follow-up schedule included laboratory tests, ultrasound, computed tomography (CT), or endoscopy. The final follow-up date was November 2021.

Assessment of the skeletal muscle index
All patients underwent routine preoperative abdominal computed tomography. As shown in Fig. 1, L3-CT images of the superior, transverse, and inferior aspects was intercepted from the Picture Archiving and Communication System. According to the EWGSOP, we used skeletal muscle area to represent skeletal muscle mass [2]. ImageJ (National Institutes of Health, Bethesda, MD, USA, 1.52v) was used to assess the skeletal muscle area using particular Hounsfield Unit (HU) criteria of -29 to 150. As needed, tissue boundaries were manually drawn for each of the three aspects. Muscle area was normalized by height squared and reported as the skeletal muscle index (SMI, cm2/m2).

Muscle strength and physical performance
Preoperative grip strength and a usual gait speed of 6 m were measured according to the EWGSOP and AWGS definitions of sarcopenia to separately determine the muscle strength and physical performance of each patient. Patients were tested for preoperative grip strength using an electronic hand dynamometer (EH101; Camry Electronics Co., Ltd., Zhongshan, Guangdong, China) with a dominant hand squeeze. The time between the first and last steps over 6 m was used to measure the usual gait speed. Both parameters were measured within 7 days preoperatively, and the maximum value of three repeated tests was recorded.

The diagnosis of sarcopenia
Patients with low skeletal muscle mass, low muscle strength, and/or low physical performance were considered to have sarcopenia, as defined by the EWGSOP and the AWGS. In this study, sarcopenia was diagnosed as follows: (1) low muscle mass (L3 skeletal muscle index (SMI) ≤ 40.8 cm2/m2 in males and ≤ 34.9 cm2/m2 in females) [2, 14]; (2) low muscle strength (grip strength < 28 kg in males and < 18 kg in females) [3]; and (3) low muscle performance (6 m usual gait speed < 1 m/s) [3]. The cutoff value was based on the inferior aspect of the L3-CT image. Sarcopenia diagnosed in the muscle area from the superior aspect of L3-CT images was referred to as superior sarcopenia, the transverse aspect of L3-CT imaging as transverse sarcopenia, and the inferior aspect of L3-CT images was considered inferior sarcopenia.

Data collection
For each patient enrolled in this study, the following data were collected at the time of patient admission: preoperative patient characteristics, including age, sex, body mass index (BMI), Charlson comorbidity index, skeletal muscle mass(SMA at the three aspects of L3-CT), muscle strength(grip strength) and physical performance (6 m usual gait speed); surgical details, including laparoscopic-assisted surgery, combined organ resection, type of resection, and operative time, were collected at the end of the patient's surgery; postoperative outcomes were collected during the postoperative hospitalization and at postoperative outpatient follow-up, including tumor pathological features and postoperative complications (within 30 days postsurgery).

Analysis
The agreement of SMA and SMI among the three aspects of L3-CT was calculated using the Wilcoxon rank sum test, and the agreement among the three diagnoses of sarcopenia was analyzed using kappa tests. Student’s t test was used to compare continuous normally distributed data, and the Mann‒Whitney U test was applied to continuous nonnormally distributed data. Spearman correlations were used for evaluation of the association between non-normally distributed data. Categorical data were compared using the chi-square test or Fisher's exact test. Univariate analysis was used to assess the relationships between categorical variables. Univariate Cox proportional hazards models with all potential baseline predictors were constructed to calculate risk ratios (HR) and 95% CI. Variables with a trend (P < 0.05) in the univariate analysis were selected, and multivariate logistic regression or Cox proportional risk models were constructed using forward stepwise variable selection. The Kaplan‒Meier method was used to estimate survival curves, and the log-rank test was used to compare the data. The data analysis was performed using the statistical package IBM SPSS Statistics software (SPSS) version 25.0.

Results

Comparison of body composition data at three L3 aspects
As shown in the Supplementary Fig. 1, we excluded unsuitable patients and finally 1116 patients were included in the study. As illustrated in Supplementary Fig. 2, a comparative analysis of the three aspects reveals that grip strength and step speed exhibit the highest correlation with SMA and SMI of the Inferior aspect. We divided all patients into two groups according to sex, and the SMA and SMI of males and females according to the three aspects are summarized in Table 1. As shown, there were significant differences in the SMA and SMI between these three dimensions for both men and women, all with increases from top to bottom. All three aspects remained highly correlated between the two aspects for both sexes (superior aspect vs. transverse aspect: SMA-R2 = 0.956; SMI-R2 = 0.935; transverse aspect vs. inferior aspect: SMA-R2 = 0.952; SMI-R2 = 0.930; inferior aspect vs. superior aspect: SMA-R2 = 0.924; and SMI-R2 = 0.887).

Diagnostic consistency of three aspects of L3
We considered inferior sarcopenia the gold standard. As shown in Table 2 and Fig. 2, the sensitivity of the superior aspect was 0.955, the specificity was 0.924, and the AUC value was 0.939. The sensitivity of the transverse aspect was 0.941, the specificity was 0.949, and the AUC value was 0.945, which suggests that the diagnosis of the transverse aspect and the superior aspect had a high degree of agreement with that of the inferior aspect. The kappa value for the transverse aspect was 0.803, and the kappa value for the superior aspect was 0.745; both factors also showed a high degree of consistency compared to that of the inferior aspect.

Population heterogeneity in three types of sarcopenia
As shown in Table 3, the prevalence of superior sarcopenia, transverse sarcopenia, and inferior sarcopenia were 19.7%, 17.4%, and 13.8%, respectively, which revealed that the superior and transverse aspects were used to screen more patients with sarcopenia than was the inferior aspect. However, there were no significant differences in hospitalization costs (superior sarcopenia vs. inferior sarcopenia, p = 0.622; transverse sarcopenia vs. inferior sarcopenia, p = 0.511), postoperative hospitalization time (superior sarcopenia vs. inferior sarcopenia, p = 0.335; transverse sarcopenia vs. inferior sarcopenia, p = 0.255), or other indicators between superior sarcopenia and transverse sarcopenia with inferior sarcopenia. Thus, we divided the patients into superior sarcopenia and transverse sarcopenia groups and found that patients with superior sarcopenia and without inferior sarcopenia, both men and women, had slightly higher SMI compared with those with inferior sarcopenia; moreover, their hospitalization cost was reduced by approximately 10%, and their overall hospitalization time was shortened by one day. Furthermore, patients with transverse sarcopenia without inferior sarcopenia had a slightly greater SMI in both men and women, compared with those with inferior sarcopenia, and while women with transverse sarcopenia without inferior sarcopenia had a slightly greater BMI than did those with inferior sarcopenia.

Short-term postoperative complications
We evaluated postoperative complications that occurred within 30 days after gastrectomy and graded the complications according to the Clavien system [24], including those of Grade II or higher. The results of the univariate and multivariate analyses of the predictors of postoperative complications are presented in Table 4. The univariate analysis showed that advanced age, superior sarcopenia, transverse sarcopenia, inferior sarcopenia, Charlson Comorbidity Index, TNM staging, combined organ removal, and open surgery were risk factors for postoperative complications. The multivariate analysis showed a higher dominance ratio for inferior sarcopenia (OR = 2.030, p < 0.001) than for superior sarcopenia (OR = 1.608, p = 0.005) and transverse sarcopenia (OR = 1.679, p = 0.004).

Long-term postoperative survival outcome
The median postoperative follow-up period was 59 months. The 5-year survival rates were 66.2%, 65.9%, and 65.5% for patients with superior, transverse, and inferior sarcopenia, respectively. As shown in Fig. 3, Kaplan‒Meier analysis revealed that overall survival (OS) (log-rank, superior sarcopenia, p = 0.0015; transverse sarcopenia, p = 0.0024; inferior sarcopenia, p = 0.003) was significantly shorter in patients with sarcopenia than in those without sarcopenia, regardless of the aspect-based diagnosis of sarcopenia. As shown in Table 5, multivariate Cox models showed that inferior sarcopenia (HR = 1.491, p = 0.004), histologic type, TNM staging, and resection type were independently associated with poorer overall survival. When using superior sarcopenia (HR = 1.408, p = 0.005) or transverse sarcopenia (HR = 1.376, p = 0.012) instead of inferior sarcopenia, the inclusion of sarcopenia remained in the multifactorial model. Compared to those of inferior sarcopenia, the risks of superior sarcopenia and transverse sarcopenia appeared to be lower.

Discussion
This study compared skeletal muscle area among three aspects of L3-CT imaging and investigated, for the first time, the differences in predicting patient prognosis caused by the diagnosis of sarcopenia using muscle data obtained from different aspects of L3-CT imaging. This study showed that the SMA and the SMI increased sequentially from the top to the bottom of the L3-CT image, SMA and SMI at the inferior aspect exhibit the most robust correlation with grip strength and walking speed. However, there was a high correlation among these three aspects, and the results were consistent with those of previous studies [25]. For this reason, more patients were diagnosed with sarcopenia in a descending sequence, from top to bottom, with more patients diagnosed with sarcopenia in the superior and transverse aspects than in the inferior aspect. Overall, there was high agreement in the diagnosis of the three aspects (superior sarcopenia vs. inferior sarcopenia, kappa value = 0.745, p < 0.001; transverse sarcopenia vs. inferior sarcopenia, kappa value = 0.803, p < 0.001). There were also significant differences in the SMA and SMI among the three aspects, with a certain consistency in diagnosis. Therefore, further analysis is needed to clarify whether this difference impacts the prediction of clinical outcomes.
Several studies have shown that patients with sarcopenia have higher hospitalization costs [26], longer hospital stays [26], and shorter postoperative survival [14] than patients without sarcopenia, in agreement with the findings of our study. We found no significant differences in postoperative length of stay and hospitalization costs among the three sarcopenia populations. After further splitting the patients into superior sarcopenia and transverse sarcopenia cohorts, patients with superior sarcopenia alone (negative for inferior sarcopenia) had a greater SMI in both sexes than did those with inferior sarcopenia. Patients in these groups also spent less on hospitalization than patients with inferior sarcopenia and had a slightly shorter length of postoperative hospitalization, while patients with transverse sarcopenia alone (negative for inferior sarcopenia) had the same SMI and postoperative hospitalization cost as patients with superior sarcopenia alone. Previous studies have revealed that a low SMI can be used as an independent risk factor for predicting postoperative length of stay [27], cost [28], complications [29] and long-term prognosis [30], whereas patients with a high SMI have been found to have a better postoperative prognosis. This may partially explain the relatively better length of stay and cost performance of patients with superior sarcopenia and transverse sarcopenia.
Further analysis of postoperative complications revealed that inferior sarcopenia had the best predictive ability (superior sarcopenia, OR = 1.608; transverse sarcopenia, OR = 1.769; inferior sarcopenia, OR = 2.030). According to our analysis of long-term survival, inferior sarcopenia appeared to retain high predictive power for survival (superior sarcopenia, HR = 1.408; transverse sarcopenia, HR = 1.376; inferior sarcopenia, HR = 1.491). Taken together, these findings showed that patients with superior and transverse sarcopenia had relatively high SMI, as described previously. This was perhaps because the superior and transverse sarcopenia cohort included more patients with suspected sarcopenia, for whom the short- and long-term prognostic performance was slightly better than that of patients with inferior sarcopenia. This has led to superior and transverse sarcopenia to present a relatively low risk predictive ability in prognostic predictions. At the root of this, despite the high correlation of SMA between the three aspects, differences remain in SMA values. We uniformly used a cutoff that was obtained according to the L3 inferior aspect as the basis for the diagnosis of low SMI in this study, and applying this cutoff to the superior aspect and transverse aspect may be the underlying cause of this difference.
This study has several limitations. First, this was a single-center study, and a larger multicenter study is needed to validate our findings. In addition, although we clarified that the cutoff value should be consistent with the intercept, whether a specific aspect of L3 selection would yield a better prognostic value remains unclear.

Conclusion
In this study, we explored in detail the diagnosis of low muscle mass in patients with sarcopenia and found that when using a uniform cutoff at the inferior aspect, it may be possible that a lower SMI at the superior and transverse aspects, compared to the inferior aspect screened out more critical patients with suspected sarcopenia, which is clearly detrimental to the predictive power of the model. We recommend that when diagnosing low SMI, the aspect of interception should be uniform and consistent with the aspect of the truncation values. Since it has the potential to be incorporated into risk-scoring systems for postoperative prognosis and to improve clinical decision-making in patients, this study of patients with gastric cancer emphasizes the need for a standardized assessment of sarcopenia.

Supplementary Information