3.1 Baseline characteristics of study sample
Of 2811 patients with stage I–IIIA lung cancer and treated with curative surgical resection, we excluded 360 patients whose preoperative CT scan images were incomplete, 72 patients who were lost to follow-up, 6 patients with neoadjuvant therapy, 183 patients with a history of other malignancies and 80 patients whose interval between the last pre-surgery chest CT examination and the operation was more than 30 days (Supplementary Fig.S1). Finally, 2110 patients were included in this study. Among the 2110 patients included, 1125 (53.32%) were males and 985 (46.68%) were females, with a median (IQR) age 59.00 (52.00–66.00) years. The PMI was found to be higher in male patients compared to female patients, with median (IQR) PMI values of 11.68 (9.99, 13.74) cm2 /m2 for males and 8.84 (7.48, 10.51) cm2 /m2 for females (Table 1).
Table 1: Demographic characteristics of study sample.
Variable
|
Patients(n =2110)
|
Age(y)a
|
59.00 (52.00, 66.00)
|
Age category
|
|
<65 years
≥65years
|
1482 (70.24)
628 (29.96)
|
Gender
|
|
Male
Female
|
1125 (53.32)
985 (46.68)
|
BMI(kg/m2 )a
|
22.27 (20.76, 24.61)
|
BMI category
|
|
<18.5kg/m2, underweight
18.5-25.0kg/m2, normal weight
≥25.0kg/m2, above normal weight
|
1566 (74.22)
110 (5.21)
434 (20.57)
|
Smoking history
|
|
Never smoker
Current or former smoker Unknown
|
1265 (59.95)
837 (39.67)
8 (0.38)
|
Tumor stage
|
|
I stage
II stage
IIIA stage
|
1325 (62.80)
374 (17.73)
411 (19.48)
|
Histologic type
|
|
Adenocarcinoma
Non-adenocarcinoma
|
1656 (78.48)
454 (21.52)
|
Surgical approach
|
|
Thoracotomy
Thoracosocope
|
762 (36. 11)
1348 (63.89)
|
Emphysema
|
|
Yes
No
Unknown
|
145 (6.87)
1964 (93.08)
1 (0.05)
|
Pleural invasion
|
|
Yes
No
Unknown
|
354 (16.78)
1750 (82.94)
6 (0.28)
|
Adjuvant chemotherapy
|
|
Yes
No
|
895 (42.42)
1215 (57.58)
|
PMA (cm2)a
|
|
All patients
Male
Female
|
27.41 (21.87, 34. 14)
32.74 (28.11, 38.57)
22.11 (18.66, 26.09)
|
PMI (cm2/m2)a
|
|
All patients
Male
Female
|
10.35 (8.46, 12.49)
11.68 (9.99, 13.74)
8.84 (7.48, 10.51)
|
SMA (cm2)a
|
|
All patients
Male
Female
|
114.90 (96.42, 139.30)
137.60 (123.45, 151.20)
96.31 (86.80, 105.95)
|
SMI (cm2/m2)a
|
|
All patients
Male
Female
|
43.72 (37.72, 49.97)
38.53 (34.75, 42.85)
48.79 (43.88, 54.20)
|
Note: Except where indicated, data are numbers of patients, with percentages in parentheses. BMI = body mass index, PMA = pectoralis muscle area, PMI = pectoralis muscle index, SMA = pectoralis muscle area , SMI = skeletal muscle index.
a Numbers are medians, with interquartile ranges in parentheses.
3.2 Relationship between pectoralis muscle index and DMFS
Using X-tile software to find the optimal cutoff values of PMI and dividing PMI into high PMI and low PMI groups (Table 2). Finally, there were 889 (90.25%) females and 853 (75.82%) males in the high PMI group.
Kaplan-Meier analysis showed low PMI is associated with lower distant metastasis-free survival rate, lung metastasis-free survival (MFS) rate, liver MFS rate, bone MFS rate, brain MFS rate and adrenal MFS rate, which log-rank test P values are all less than 0.001 (Figure 1.A-F). Univariate Cox regression analysis revealed that PMI was associated with DMFS and the metastasis-free survival at five specific metastatic sites, all with P < 0.001. In the multivariable analysis, PMI was the independent predictor for the DMFS (model 1: hazard ratio [HR] = 1.25, 95% CI: [1.02, 1.52], P = 0.03; model 2: HR = 1.35, 95% CI: [1.11, 1.65], P = 0.003), lung MFS (model 1: HR = 1.36, 95% CI: [1.11, 1.68], P = 0.004; model 2: HR = 1.47, 95% CI: [1.19, 1.81]; P < 0.001) and bone MFS (model 1: HR = 1.26, 95% CI: [1.01, 1.56], P = 0.04; model 2: HR = 1.38, 95% CI: [1.11, 1.73]; P = 0.004) (Table 3).
Table 2: Segmentation of pectoralis muscle area on CT images and the sex-specific cut-off values of pectoralis muscle index.
Segmentation of pectoralis muscle area on CT images
|
|
|
Gender
Cut-off values
|
Male
|
Female
|
Pectoralis muscle index (cm2/m2)
|
9.90
|
6.30
|
Note:(A) The pectoralis muscle area was measured at the level of the 4th thoracic vertebra transverse process. (B) Segmentation of the pectoralis major and minor muscle (red area).
Table 3: Univariable and multivariable analyses of pectoralis muscle index for metastasis-free survival.
Variable
|
Event
|
Univariate analysis
|
Multivariate analysis(M1)a
|
Multivariate analysis(M2)b
|
HR(95% CI)
|
P.value
|
HR(95% CI)
|
P.value
|
HR(95% CI)
|
P.value
|
DMFS
|
625 (29.62%)
|
1.45 (1.20, 1.76)
|
<0.001
|
1.25 (1.02, 1.52)
|
0.03
|
1.35 (1.11, 1.65)
|
0.003
|
Lung MFS
|
540 (25.59%)
|
1.66 (1.36, 2.02)
|
<0.001
|
1.36 (1.11, 1.68)
|
0.004
|
1.47 (1.19, 1.81)
|
<0.001
|
Liver MFS
|
466 (22.09%)
|
1.57 (1.27, 1.95)
|
<0.001
|
1.22 (0.97, 1.53)
|
0.09
|
1.31 (1.05, 1.66)
|
0.02
|
Bone MFS
|
494 (23.41%)
|
1.56 (1.27, 1.92)
|
<0.001
|
1.26 (1.01, 1.56)
|
0.04
|
1.38 (1.11, 1.73)
|
0.004
|
Brain MFS
|
518 (24.55%)
|
1.48 (1.20, 1.81)
|
<0.001
|
1.20 (0.96, 1.49)
|
0.10
|
1.32 (1.06, 1.64)
|
0.01
|
Adrenal MFS
|
470 (22.27%)
|
1.55 (1.25, 1.92)
|
<0.001
|
1.21 (0.96, 1.51)
|
0.10
|
1.31 (1.04, 1.66)
|
0.02
|
Note: HR = hazard ratio, CI = confidence interval, DMFS = distant metastasis-free survival, MFS = distant metastasis-free survival. a Multivariate analysis (M1) was adjusted for age category, gender, BMI category.
b Multivariate analysis (M2) was adjusted for multivariate analysis (M1) plus smoking history, tumor stage, histologic stage, surgical approach, emphysema, pleural invasion and adjuvant chemotherapy.
3.3 Relationship between pectoralis muscle index and RFS
Kaplan-Meier analysis demonstrated a reduced recurrence-free survival rate and a reduced overall survival (OS) rate in patients with a low PMI, with all P values less than 0.001 (Figure 1.G-H). Univariable Cox regression analysis revealed that PMI was associated with patients’ RFS. In the multivariable analysis, PMI (model 1: HR = 1.24, 95% CI: [1.02, 1.50], P = 0.03; model 2: HR = 1.34, 95% CI: [1.10, 1.62], P = 0.004) remained an independent predictive parameter of RFS (Table 4). In addition, we discussed the relationship between PMI and OS. The results of univariable Cox proportional hazards regression analyses showed a correlation between PMI (HR = 1.58, 95% CI: [1.28, 1.96], P < 0.001) and OS. After adjusting for other confounding factors, a strong correlation between PMI and OS persisted (model 1: HR = 1.22, 95% CI: [0.97, 1.53], P = 0.08; model 2: HR = 1.33, 95% CI: [1.06, 1.68], P = 0.01) (Supplementary Table S1).
Table 4 : Univariable and multivariable analyses of clinical and pectoralis muscle index for recurrence-free survival.
Variable
|
Univariate analysis
|
Multivariate analysis(M1)a
|
Multivariate analysis(M2)b
|
HR(95% CI)
|
P.value
|
HR(95% CI)
|
P.value
|
HR(95% CI)
|
P.value
|
Age category
|
|
|
|
|
|
|
<65 years
|
Ref
|
|
|
|
Ref
|
|
≥65 years
|
1.23 (1.05, 1.44)
|
0.01
|
|
|
1.07 (0.90, 1.26)
|
0.45
|
Gender
|
|
|
|
|
|
|
Female
|
Ref
|
|
|
|
Ref
|
|
Male
|
1.59 (1.36, 1.85)
|
<0.001
|
|
|
1.39 (1.10, 1.76)
|
0.01
|
BMI category
|
|
|
|
|
|
|
18.5-25.0kg/m2, normal weight
|
Ref
|
|
|
|
Ref
|
|
<18.5kg/m2, underweight
|
1.46 (1.08, 1.97)
|
0.01
|
|
|
1.44 (1.06, 1.97)
|
0.02
|
≥25.0kg/m2, above normal weight
|
0.97 (0.80, 1. 17)
|
0.75
|
|
|
1.04 (0.85, 1.26)
|
0.70
|
Smoking history
|
|
|
|
|
|
|
Never smoker
|
Ref
|
|
Ref
|
|
Ref
|
|
Current or former smoker
|
1.41 (1.21, 1.65)
|
<0.001
|
1.02 (0.82, 1.27)
|
0.86
|
0.80 (0.65, 1.01)
|
0.06
|
Tumor stage
|
|
|
|
|
|
|
I stage
|
Ref
|
|
Ref
|
|
Ref
|
|
II stage
|
3.06 (2.52, 3.71)
|
<0.001
|
2.91 (2.39, 3.54)
|
<0.001
|
2.57 (2.07, 3. 19)
|
<0.001
|
IIIA stage
|
4.47 (3.74, 5.33)
|
<0.001
|
4.30 (3.60, 5. 14)
|
<0.001
|
3.67 (2.99, 4.51)
|
<0.001
|
Histologic type
|
|
|
|
|
|
|
Adenocarcinoma
|
Ref
|
|
Ref
|
|
Ref
|
|
Non-adenocarcinoma
|
1.64 (1.38, 1.94)
|
<0.001
|
1.40 (1.17, 1.68)
|
<0.001
|
0.85 (0.70, 1.03)
|
0.10
|
Emphysema
|
|
|
|
|
|
|
Yes
|
Ref
|
|
Ref
|
|
Ref
|
|
No
|
0.74 (0.56, 0.98)
|
0.04
|
0.88 (0.66, 1. 16)
|
0.36
|
0.80 (0.60, 1.06)
|
0.11
|
Surgical approach
|
|
|
|
|
|
|
Thoracotomy
|
Ref
|
|
Ref
|
|
Ref
|
|
Thoracoscope
|
0.41 (0.35, 0.47)
|
<0.001
|
0.44 (0.37, 0.51)
|
<0.001
|
0.67 (0.56, 0.80)
|
<0.001
|
Adjuvant chemotherapy
|
|
|
|
|
|
|
Yes
|
Ref
|
|
Ref
|
|
Ref
|
|
No
|
0.57 (0.49, 0.66)
|
<0.001
|
0.59 (0.50, 0.69)
|
<0.001
|
0.91 (0.77, 1.07)
|
0.26
|
Pleural invasion
|
|
|
|
|
|
|
Yes
|
Ref
|
|
Ref
|
|
Ref
|
|
No
|
0.72 (0.60, 0.87)
|
0.001
|
0.76 (0.63, 0.92)
|
0.005
|
0.90 (0.74, 1. 10)
|
0.30
|
PMI (cm2/m2)
|
|
|
|
|
|
|
High
|
Ref
|
|
Ref
|
|
|
|
Low
|
1.44 (1.20, 1.73)
|
<0.001
|
1.24 (1.02, 1.50)
|
0.03
|
1.34 (1.10, 1.62)
|
0.004
|
Note: BMI = body mass index, HR = hazard ratio, CI = confidence interval, PMI = pectoralis muscle index.
a Multivariate analysis (M1) was adjusted for age category, gender, BMI category.
b Multivariate analysis (M2) was adjusted for multivariate analysis (M1) plus smoking history, tumor stage, histologic stage, surgical approach, emphysema, pleural invasion and adjuvant chemotherapy.
3.4 Subgroup analysis
The results of the subgroup analysis showed that, for the four endpoints, patients with low PMI had a higher hazard ratio in most subgroups, similar to the general population. The baseline characteristics of the patients and the PMI did not show significant interaction (all P > 0.05) (Figure 2).
3.5 Gradient boosting model for comparison of parameters
The factors with P < 0.05 obtained in model 1 of the multivariate Cox regression analysis were included in the gradient boosting model (GBM) analysis (Table 4, Supplementary Table S2). GBM demonstrated that the impact of PMI on RFS, lung metastasis-free survival and bone metastasis-free survival was second only to tumor stage and surgical approach, and was higher than BMI (Figure 3).