*2.3. Measurements*

All of the following measurements were performed on the same day for each participant.

#### 2.3.1. Trunk Muscle Mass Measurement by BIA (TMM–BIA)

We measured the trunk muscle mass (kg) of participants using the BIA method with a body composition analyzer (MC-780A, Tanita Co., Tokyo, Japan). BIA is a non-invasive examination technique used to determine body composition by measuring the electrical resistance (bioimpedance) of living tissues [21]. The BIA device (MC-780A) measures bioimpedance using six electrical frequencies (1, 5, 50, 250, 500, and 1000 kHz). It can accurately identify bone and fat because it distinguishes tissues by their bioimpedance. Muscle mass (kg) was calculated by subtracting fat mass and bone mass from the total body weight (kg). Furthermore, trunk muscle mass (kg) was calculated by subtracting the ASM (kg) from the muscle mass of the whole body (kg).

**Figure 1.** Flowchart of the included and excluded participants. A total of 380 participants were enrolled for this analysis.

#### 2.3.2. Quantitative Evaluation of Trunk Muscle on MRI

In this study, MRI evaluations were performed using the Achieva 3.0 Quasar (Koninklijke Philips N.V., Amsterdam, Netherlands). A T2-weighted axial image (TR = 7670, TE = 90, FOV = 170 × 170 mm, slice = 5 mm) was used to measure the CSA of PVM, including ES, MF, and PM, at the L3/4 level, using the "pencil tool" from the 32-bit OsiriX software (version 3.8.1, Pixmeo, Geneva, Switzerland). The CSA including infiltrated fat was measured and determined, and then the intramuscular fat based on regions of interest (ROIs) with intensity changes was differentiated. The fat-free CSA for each PVM was then calculated as the difference between these two values [22].

#### 2.3.3. Functional Evaluation of Trunk Muscles

The BMS of each participant was determined by measuring the maximal isometric strength of the trunk muscles in a standing position with 30◦ of lumbar flexion using a digital BMS meter (T.K.K.5402, TAKEI, Niigata, Japan) [10,11]. After performing warm-up exercises called "radio calisthenics", the participants underwent the BMS measurement twice. The average force from two trials was recorded. As the minimum measurable value of the digital BMS meter is 20 kg, in case the participant's BMS was too weak to be measured, it was not recorded and was excluded from the analysis of BMS.

#### *2.4. Statistical Analysis*

We investigated the correlation between TMM–BIA and the CSA of PVM and fat-free CSA of PVM using Spearman's rank correlation coefficient. The relationship between TMM–BIA and the CSA of each individual PVM (ES, MF, and PM) was also evaluated. Additionally, we examined the association between TMM–BIA and BMS. Patient demographics were compared using Student's *t*-tests. All statistical analyses were performed using Statistical Package for the Social Sciences (SPSS Inc., version 19.0, Chicago, IL, USA). Correlation strengths were categorized as very weak (<0.20), weak (0.20–0.39), moderate (0.40–0.59), strong (0.60–0.79), or very strong (≥0.80). Statistical significance was set at *p* < 0.05.

## **3. Results**

Data from 380 participants in the Shiraniwa study (152 males, 228 females; mean age, 73.4 years) who underwent TMM–BIA, lumbar MRI, and BMS measurements were analyzed in this study. The participants' characteristics are summarized in Table 1.


**Table 1.** Characteristics of the participants of the Shiraniwa study.

Data are presented as mean (standard deviation). BMI, body mass index; TMM–BIA, trunk muscle mass measured by bioelectrical impedance analysis; CSA, cross-sectional area; PVM, paravertebral muscles. Student's *t*-test was used to compare groups.

> A significant and strong correlation was found between TMM–BIA and the CSA of PVM (r = 0.746, *p* < 0.01) (Figure 2), and between TMM–BIA and the fat-free CSA of PVM (r = 0.807; *p* < 0.01) (Figure 3). Similarly, TMM–BIA was significantly correlated with the CSA of each individual PVM (Figure 4). The CSA of PM was strongly correlated with the TMM–BIA (fat included, r = 0.752, *p* < 0.01; fat-free, r = 0.766, *p* < 0.01), whereas the CSA of MF, the smallest muscle of the PVM, was moderately correlated with TMM–BIA (fat included, r = 0.439, *p* < 0.01; fat-free, r = 0.571, *p* < 0.01). In addition, the CSA of ES was moderately correlated with TMM–BIA (fat included, r = 0.554, *p* < 0.01; fat-free, r = 0.658, *p* < 0.01) (Table 2). TMM–BIA and BMS were strongly correlated (r = 0.726, *p* < 0.001), although the strength of some participants could not be measured due to back pain (Figure 5).

**Figure 2.** Correlation between TMM–BIA and the CSA of PVM. There was a significant correlation between TMM–BIA and the CSA of PVM with r = 0.746. TMM–BIA, trunk muscle mass measured by bioelectrical impedance analysis; CSA, cross-sectional area; PVM, paravertebral muscles.

**Figure 3.** Correlation between TMM–BIA and the CSA of PVM without intramuscular fat. There was a significant correlation between TMM–BIA and the CSA of fat-free PVM with r = 0.807. TMM–BIA, trunk muscle mass measured by bioelectrical impedance analysis; CSA, cross-sectional area; PVM, paravertebral muscles.

**Figure 4.** Correlations between TMM–BIA and the CSA of each individual PVM (upper row, total; lower row, excluding intramuscular fat). The CSA of the PM showed a strong correlation with the TMM–BIA (total, r = 0.752; fat-free, r = 0.766), whereas the CSA of MF, the smallest muscle of the PVM, showed a moderate correlation (total, r = 0.439; fat-free, r = 0.571). In addition, the CSA of ES had a moderate to strong correlation to the TMM–BIA (total, r = 0.554; fat-free, r = 0.658), respectively. TMM–BIA, trunk muscle mass measured by bioelectrical impedance analysis; CSA, cross-sectional area; PVM, paravertebral muscle; ES, erector spinae; MF, multifidus; PM, psoas major.


**Table 2.** Correlations between TMM–BIA and each PVM with and without intramuscular fat.

Data are presented as mean (standard deviation). TMM–BIA, trunk muscle mass measured by bioelectrical impedance analysis; CSA, cross-sectional area; PVM, paravertebral muscles; R, correlation coefficient; ES, erector spinae; MF, multifidus; PM, psoas major.

**Figure 5.** Correlation between TMM–BIA and back muscle strength. There was a strong correlation of r = 0.726, even though some of the participants exhibited minimum strength because of pain. TMM–BIA, trunk muscle mass measured by bioelectrical impedance analysis.

#### *3.1. Case Presentation*

#### 3.1.1. Case 1

A 67-year-old male with a history of hepatitis and diabetes mellitus had a high TMM–BIA of 31.5 kg. The CSA of PVM on the MRI was 75.4 cm<sup>2</sup> (fat included) and 68.04 cm<sup>2</sup> (fat-free) (Figure 6). He reported his low back pain as 0 mm on a visual analog scale. His back muscle strength was 85.5 kg.

#### 3.1.2. Case 2

A 70-year-old female with a history of diabetes mellitus and osteoporosis had a low TMM–BIA volume of 14.2 kg. MRI showed severe muscular atrophy and fatty degeneration in her PVM (Figure 7). The CSA was 36.59 cm<sup>2</sup> (fat included) and 20.02 cm<sup>2</sup> (fat-free). She reported severe low back pain as 76 mm on a visual analog scale. Her back muscle strength was too weak to be recorded (less than 20 kg).

**Figure 6.** Case presentation 1. A 67-year-old male with no symptoms of low back pain had a CSA of PVM and fat-free PVM of 75.4 and 68.04 cm2, respectively. The fat-free percentage of PVM was 90.2%. CSA, cross-sectional area; PVM, paravertebral muscles.

**Figure 7.** Case presentation 2. A 70-year-old female with severe low back pain. The patient's CSA of PVM and fat-free PVM was 36.59 and 20.02 cm2, respectively. The fat-free percentage of PVM was 54.7%. TMM–BIA, trunk muscle mass measured by bioelectrical impedance analysis; CSA, cross-sectional area; PVM, paravertebral muscles.
