Background

Obesity is a well-known causative and prognostic factor of breast cancer [1, 2]. Body mass index (BMI), which incorporates height and weight, is the representative index of obesity, and only falls short as a well-rounded index because it fails to represent body composition. The clinical importance of BMI has been studied and proven for diseases other than breast cancer, such as diabetes mellitus, chronic metabolic disease, and others. However, in the case of chronic metabolic diseases, it has been shown that low muscle volume is adversely prognostic of the disease, rather than obesity (specifically, high BMI), suggesting a prognostic value for not only fat but also muscle. Accordingly, many studies have investigated the potential mechanistic role of skeletal muscle in catabolism and in disease pathology [3, 4].

Skeletal muscle occupies a substantial volume of the body, and a loss in skeletal muscle leads to poorer outcome in many chronic diseases [5, 6]. Cachexia, for instance, in cancer patients is associated with morbidity and with mortality [7,8,9]. The importance of protein, the main constituent of muscle, has been emphasized for chronic diseases, such as cancer and chronic obstructive pulmonary disease [10,11,12]. Increasing the mass of skeletal muscle in obese patients through changes in lifestyle habits such as increased exercise is also associated with improved prognosis in chronic diseases.

With the development of imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI), we are now able to obtain a cross-sectional image of any body part and evaluate its composition. Thus, analysis of the patient’s body composition is now becoming increasingly reliable and consistent. Reliability and consistency have been achieved through the identification of regions of the body where cross-sectional images can represent the entire organ or tissue, which can be used uniformly across examinations. Several studies have investigated the most representative region within the body for skeletal muscle volume, and have found that taking a cross-sectional CT image of the lumbar skeletal muscle, specifically at the 3rd vertebrae, provides this information [13,14,15,16,17,18].

Studies on the prognosis of breast cancer that are based on skeletal muscle volume are lacking [19]. Many studies have conclusively demonstrated the association between obesity and the incidence and prognosis of breast cancer; however, studies showing the association with quantitatively measured skeletal muscle are sparse, and only weakly demonstrate that relationship [1]. In this study, we investigated, using CT images of the chest to derive tissue volumes, whether the prognosis of breast cancer is affected by the patient’s muscle or fat volume. We also investigated whether the prognosis of breast cancer differs according to skeletal muscle volume among patients with similar BMIs.

Methods

Study population

We analyzed the medical records of 3909 consecutive patients with a confirmed diagnosis of breast cancer between January 2001 and December 2009 at the National Cancer Center (NCC), Republic of Korea. The inclusion criteria consisted of having received a chest CT within 2 years of diagnosis, chest CT slices of 10 mm in thickness, and CT images that include the 3rd lumbar vertebrae. The patients without adequate CT image (n = 2414), male breast cancer patients (n = 4), patients with stage IV breast cancer (n = 15), who has a final pathology of carcinoma in situ, breast sarcoma, or malignant phyllodes tumor (n = 6), a previous history of breast cancer patients (n = 10) were excluded.

Data were extracted from the electronic medical record system at NCC to identify the following patient information: age at the time of diagnosis, height, weight, BMI at the time of diagnosis, stage of breast cancer, pathological characteristics, date of last clinic visit, and death records. We generally used the pathological stage, but used the clinical stage for patients who underwent neoadjuvant chemotherapy. The number of patients who underwent neoadjuvant chemotherapy was 186 (12.4%). The information collected was retrospectively analyzed in this study. The study was reviewed and approved by the NCC’s Institutional Review Board of Ethics (IRB), and the requirement of written informed consent was waived (IRB No.: NCC2015-0006).

Evaluation of muscle volume and fat volume on CT

The total cross-sectional area of skeletal muscle on total 10-mm-thick CT slices taken at the 3rd lumbar vertebrae (Fig. 1) was measured by a radiology specialist (Lee) [12, 14, 15]. We derived the skeletal muscle volume encompassing the area of the CT slice just below the intervertebral disc between L2 and L3. Using the same CT image, we measured fat volume in the same manner. Visceral adipose tissue and subcutaneous adipose tissue were not distinguished when measuring fat volume. Images were analyzed using the Infinitt healthcare Xelis 3D program (Infinitt, Seoul, South Korea, version BN 10). We measured the cross-sectional areas by Hounsfield unit (HU), with skeletal muscle from + 100 to -30 and adipose tissue from − 250 to − 50.

Fig. 1
figure 1

Images of cross-sectional area of skeletal muscle a with the Hounsfield unit (HU) from + 100 to − 30, and of adipose tissue b from − 250 to − 50, using the Infinitt Healthcare Xelis 3D program (Infinitt, Seoul, South Korea, version BN 10). Blue colored area is the area of interest

Full size image

Statistical analyses

The primary end-points of our study were overall survival (OS) and recurrence-free survival (RFS). In addition, we evaluated locoregional recurrence-free survival (LRFS) as a second end-point. RFS was defined as the time in years between the diagnosis date and distant metastasis or locoregional recurrence; LRFS was defined as the time in years between the diagnosis date and recurrence of ipsilateral breast, axilla, and chest wall. We defined muscle volume, fat volume, and muscle-to-fat ratio as dichotomous variables using the median as the cut point, because no benchmark values for these variables currently exist in Korean women. Thus, the variables were grouped according whether the value was higher or lower than the median. Differences in baseline characteristics stratified according to muscle volume groups, fat volume groups, and muscle-to-fat ratio groups were examined through the Chi-square and the Wilcoxon rank-sum tests as appropriate. Estimates of survival times were calculated using the Kaplan–Meier method. The log-rank test and the univariate Cox regression model were used to evaluate whether each factor was prognostic for survival. Factors selected through univariate regression were included in the multivariable Cox regression model using a backward stepwise process. Muscle volume, fat volume, and muscle-to-fat ratio were then added to the optimal model after adjusting for all other factors, to evaluate whether each factor was prognostic for our three end-points. Statistical significance was defined as P < 0.05 and the confidence interval (CI) was set at 95%. We performed all statistical analyses with STATA v.13 (StataCorp LP, Texas).

Results

Patient characteristics

We analyzed a total of 1460 patients who were eligible, according to our inclusion and exclusion criteria, from the 3909 consecutive patients whose medical data we assessed. The median age of the patients was 46.0 years (range 25–77 years). The proportions of different stages of invasive breast cancer were as follows: stage I, 38.2%; stage II, 46.5%; and stage III, 15.3% (Table 1). We found that 70.3% and 65.3% of patients were positive for the estrogen receptor (ER) and progesterone receptor (PR), and that 14.8% were positive for human epidermal growth factor receptor-2 (HER-2).

Table 1 Clinicopathologic characteristics of the patients and association with muscle volume, fat volume, and muscle-to-fat ratio
Full size table

The median follow-up period was 8.07 years (range 0.63–13.32). During follow-up, 185 patients (12.7%) experienced disease recurrence, and 93 patients (6.4%) died. The 5 year OS rate was 96.5%, the 5 year RFS rate was 92.1%, and the 5 year LRFS rate was 98.6%.

The median level of total cholesterol was 184 mg/dL (range, 106–346 mg/dL), the median level of total protein was 7.3 g/dL (range, 3.8–9.1 g/dL), and the median level of albumin was 4.4 g/dL (range, 2.6–6.1 g/dL). Patients were co-morbid with the following proportions of co-morbidities relating to metabolic syndrome: 3.1% of patients had diabetes mellitus and 10.1% had hypertension.

The median volume of skeletal muscle was 93.3 cc (Supplement Fig. 1; range 39.6–236.9 cc) and the median volume of fat, analyzed from the same cross-sectional image as that used to analyze muscle volume, was 420.1 cc (range 19.5–1392.3 cc). The median muscle-to-fat ratio was 0.22 (range 0.08–3.18).

Differences of clinicopathologic characteristics and clinical outcomes by muscle volume, fat volume, and muscle-to-fat ratio

Table 1 shows the association of clinicopathologic factors with muscle volume and fat volume. Elderly patients (≥ 50) had higher fat volume (P < 0.001). Patients with higher BMI had higher muscle volume, fat volume, and muscle-to-fat ratio (P < 0.001). Patients with higher muscle volume had higher grade tumors (P = 0.023). Fat volume and muscle-to-fat volume were associated with advanced clinical stage (P = 0.011, P = 0.003) and large tumor (P = 0.043, P = 0.001). However, hormone receptor status was not statistically correlated with muscle volume and fat volume. Muscle volume showed a positive correlation with fat volume (P < 0.001), but not with muscle-to-fat ratio (P = 0.958).

When the median was used as the cut point, patients with higher muscle volume than the median volume showed significantly more favorable outcomes than those with lower muscle volume (5 year OS rate, 98.0% vs. 94.9%, P = 0.008 and 5 year RFS rate 94.6% vs. 89.6%, P = 0.031, Fig. 2). However, muscle volume had no impact on LRFS (99.0% vs. 98.3%, P = 0.899). In addition, fat volume did not affect clinical outcomes (5-year OS rate, 96.8% vs. 96.1%, P = 0.629, 5-year RFS rate 92.1% vs. 92.1%, P = 0.463, and 5-year LRFS 98.3% vs. 99.0%, P = 0.290).

Fig. 2
figure 2

Overall survival curves and recurrence-free survival curves of breast cancer patients by muscle volume (a, b) and fat volume (c, d)

Full size image

Univariate and multivariable analysis for rates of OS and RFS

In the univariate analysis, we found that the following factors were statistically significant for OS: patient’s age at diagnosis, stage including T stage and N stage, tumor grade, ER, PR, Ki-67, type of operation, anti-hormone therapy, and muscle volume (Table 2). After adjusting for these factors in the multivariate model, we found that muscle volume remained a significant, strong prognostic variable for survival (HR 0.56, 95% CI 0.34–0.92, P = 0.022).

Table 2 Univariate and multivariate analyses on overall survival and on recurrence-free survival
Full size table

Statistically significant prognostic factors for RFS were stage, PR, type of operation, radiotherapy, anti-hormone therapy, and muscle volume. In multivariate analysis, muscle volume was an independent factor significantly associated with recurrence (HR 0.72, 95% CI 0.52–0.99, P = 0.046).

Subgroup analysis for clinical impact of muscle volume

Prognosis in breast cancer patients differs according to intrinsic subtypes and stage at diagnosis; we therefore analyzed the association with muscle volume in each subtype. In the triple-negative breast cancer group, those with higher muscle volume showed better OS and RFS (HR 0.33, 95% CI 0.14–0.77 for OS; HR 0.34, 95% CI 0.16–0.72 for RFS, Supplement Fig. 2a). However, there were no statistical differences in OS and RFS between those with higher muscle volume and lower muscle volume. In subgroup analysis for stage at diagnosis, those with higher muscle volume in stage III showed better OS and RFS (HR 0.49, 95% CI 0.26–0.93 for OS; HR 0.53, 95% CI 0.32–0.89, for RFS, Supplement Fig. 2b), but not in those with stage I or II.

In subgroup analyses with respect to different BMIs and ages, muscle volume was a significant prognosticator for OS in patients with normal BMI and those with higher BMI (HR 0.55, 95% CI 0.32–0.93, P = 0.034 in BMI < 25.0; HR 0.44, 95% CI 0.21–0.91, P = 0.026 in BMI ≥ 25.0, Table 3). In addition, older patients (≥ 50) with higher muscle volume showed better OS and RFS (HR 0.44, 95% CI 0.23–0.85, P = 0.015 for OS; HR 0.55, 95% CI 0.34–0.90, P = 0.017 for RFS). However, fat volume was not a prognostic factor, regardless of BMI and age.

Table 3 Prognostic impact of muscle volume and fat volume according to BMI and age group
Full size table

Discussion

Obesity is a well-known causative and prognostic factor in breast cancer [1, 2]. Several studies have investigated the role of obesity and the associated microenvironment that promotes obesity in obese patients. With the development of imaging modalities such as CT and MRI, these techniques can now be used to differentiate body composition accurately, and to quantitatively assess the different tissue types within a specific area of interest. Such three-dimensional volumetric imaging techniques are gaining focus from a healthcare perspective, because information derived from such techniques can be utilized to make disease-specific evaluations and recommendations. For instance, a study on patients with sarcopenic obesity recruited from a single institution found that the prognosis in patients with less muscle mass, determined using CT and MRI, is poorer than those with more muscle mass. Consequently, a recommendation for sarcopenically obese patients based on this study is to implement changes in diet, lifestyle, or care that would promote build-up of muscle mass.

Here, we report the results of a retrospective study on a large cohort of breast cancer patients recruited from a single institution. We quantitatively measured skeletal muscle volume and fat volume based on patients’ CT images, which were 1 cm in thickness and taken at the intervertebral disc between the 2nd and 3rd lumbar vertebra. Our optimized regression model showed that the amount of muscle was significantly correlated with OS, RFS, and LRFS, after adjusting for other significant factors. In particular, we found that having a high muscle-to-fat ratio was more prognostic for favorable survival than having a lower ratio. Using the multivariable Cox regression model, muscle volume and muscle-to-fat ratio was demonstrated to have a positive impact on prognosis, especially on RFS. But fat volume alone did not show a correlation with survival, even with relapse of the disease.

There are a number of limitations to this study. As our study was undertaken retrospectively, we found that two variables were inconsistent and uncontrolled among patients: the time between the date of diagnosis and the date of chest CT; and the diverse imaging protocol such as different slice thickness. A large number of patients had to be excluded because they did not fit the inclusion criteria for these variables. So, selection bias may have arisen during the process of patient recruitment. Second, we used CT image for evaluating muscle and fat volume instead of MRI imaging. Although CT has relatively short time for image acquisition, low cost compared with MRI, MRI has been widely used in these days because it is more useful for physiologic imaging, diagnosis, and prediction of prognosis [20,21,22,23]. Third, important factors related to prognosis of the disease, such as the type of chemotherapy, radiation therapy, and anti- hormonal therapy, were not analyzed in this study. Lastly, the written records of the patients were made several years ago; therefore some data, such as immunohistochemistry staining, were unavailable. Despite the limitations, the study deserves merit for highlighting a strong, positive correlation between muscle volume and breast cancer prognosis in 1494 breast cancer patients recruited from a single institution. The results of this study are anticipated to provide evidence for the necessity of increasing muscle mass by adopting a healthier lifestyle for cancer patients to enhance prognosis. The limitations of the present study may be overcome by investigating the impact of muscle volume on survival not only in cancer patients, but also in patients with muscle wasting disease. Further, a parallel study investigating the effect of muscle volume on prognosis in patients treated with other anti-cancer treatments would improve the qualitative findings of this study.

Conclusion

This study showed that breast cancer patients with higher skeletal muscle volumes showed more favorable prognosis regardless of BMI. In addition, muscle volume affects clinical outcomes in older patients.