Computed tomography based analyses of body mass composition in HER2 positive metastatic breast cancer patients undergoing first line treatment with pertuzumab and trastuzumab

Body composition parameters (BCp) have been associated with outcome in different tumor types. However, their prognostic value in patients with HER2-positive metastatic breast cancer (BC) receiving first line treatment with dual anti-HER2 antibody blockade is unknown. Preclinical evidences suggest that adipocytes adjacent to BC cells can influence response to anti-HER2 treatments. We retrospectively analyzed Computed Tomography (CT)-based BCp from 43 patients with HER2-positive metastatic BC who received first line pertuzumab/trastuzumab-based treatment between May 2009 and March 2020. The impact of baseline CT-based BCp on progression-free survival (PFS) was tested using Kaplan–Meier estimates and univariate and multivariate Cox regression models. We found a significantly worse PFS for patients with high baseline subcutaneous fat index (median 7.9 vs 16.1 months, p = 0.047, HR = 2.04, 95%CI 1–4.17) and for those with high total abdominal fat index (8.1 vs 18.8 months, p = 0.030, HR = 2.17, 95%CI 1.06–4.46). Patients with baseline sarcopenia did not show shorter PFS compared to those without sarcopenia (10.4 vs 9.2 months, p = 0.960, HR = 0.98, 95%CI 0.47–2.03). Total abdominal fat index remained a significant predictor of PFS at multivariate analysis. Our findings suggest that a high quantity of total abdominal fat tissue is a poor prognostic factor in patients receiving trastuzumab/pertuzumab-based first-line treatment for HER2-positive metastatic BC.

At the time of diagnosis, BC stage was I to III in 25 cases (58.1%) and IV in 18 cases (41.9%). ER at diagnosis were positive in 28 cases (65.1%) and PgR in 13 cases (30.2%). Out of 43 patients, 33 (76.7%) were in menopause. All patients received taxane based chemotherapy (paclitaxel, twelve administrations at a dose of 80 mg/ m 2 every 7 days, or docetaxel six cycles at a dose of 75 mg/m 2 every 21 days) and trastuzumab plus pertuzumab for six cycles.
All hormone-receptor positive patients received as maintenance endocrine treatment (plus dual anti-HER2 antibody blockade) in according to their menopausal status; in particular, ten of these underwent to exemestane plus luteinizing hormone-releasing hormone (LHRH) analogs and 33 letrozole, based on physician's choice. At the beginning of the therapy, the ECOG performance status was 0 in 39 patients (90.7%) and from 1 to 2 in 4 patients (9.3%). Twenty-one patients (48.8%) had normal BMI, 16 were overweight (25 ≤ BMI < 30, 37.2%) and 6 were obese (BMI ≥ 30, 14%). Fifteen patients (34.9%) demonstrated baseline sarcopenia whereas 28 patients (65.1%) had a normal skeletal muscle index. The agreement between the two readers on the BCp measurements was excellent (0.998 95% CI 0.997-0.998, where 0 is no agreement and 1 is perfect agreement).
All additional patient characteristics and body composition parameters are described in Table 1.
Univariate analysis. The  Among the clinical parameters, expression of ER positively affects PFS (9.7 vs 6.2 months, p = 0.035, HR = 0.31, 95% CI 0.10-0.98). All the other clinical and metabolic parameters considered, such as age, Ki67, LDL, menopausal status and sites of metastasis, did not show any significant correlation with PFS. Result of the univariate analysis are summarized in Table 2.
High subcutaneous fat index (p = 0.013, HR = 4.57, 95% CI 1.38-15.08) and high total abdominal fat index (p = 0.002, HR 8.56, 95% CI 2.19-33.47) demonstrated an even higher association with reduction of PFS in the subgroup of patients with the novo metastatic disease (Table 3, stage IV), whereas no significant effect was found in patients with stage I-III at diagnosis. Interestingly, Visceral Fat Index showed a significant qualitative interaction with stage (p = 0.024), with an opposite effect on PFS, in patients with stage I-III compared to those with de novo stage IV disease, that deserves further investigation. While the interaction between subcutaneous fat index and stage was not significant (p = 0.1), high total abdominal fat index showed a significant quantitative interaction with stage (p = 0.026). Multivariate and multiparametric analysis. At multivariate analysis, that included ER, SFI and TAFTI as significant univariate predictors and menopausal status and BMI as clinically relevant variable, only ER, TAFTI and menopausal status showed an independent significant association with PFS: high total abdominal fat index as a detrimental factor (p = 0.05, HR = 3.47, 95% CI 1.00-13.04), ER-positivity and post-menopausal status as protective factors (p = 0.017, HR = 0.33, 95% CI 0.14-0.82, and p = 0.021, HR = 0.36, 95% CI 0.15-0.86, respectively) (Table 4). Additionally, multivariate analysis was performed for the OS, but no significant correlation were found (table included in the supplementary materials).
Multiparametric analysis demonstrated a strong correlation between BMI and TAFTI, SFI and VFI (p < 0.0001), as well as with SMI (p = 0.005). Additionally, a strong correlation was found between ER and PgR (p < 0.0001) and between weight and most of the body composition parameters (Table 5).

Discussion
Breast cancer is a heterogeneous disease, whose development entails the accumulation of various genetic and epigenetic alterations, ultimately leading to deregulated intracellular signaling. However, is currently evident that factors such as excess body weight have an important promoting role 13,14 .
In the early breast cancer setting, a significant amount of data indicates that overweight (BMI 25-29.9 kg/m 2 ) or obesity (BMI ≥ 30 kg/m 2 ) are associated with shorter survival 5,15 . On the contrary, in metastatic BC patients, the available evidence is scant and discordant [16][17][18][19][20] . The evidence is even more inconsistent when the various molecular subtypes are considered. To the best of our knowledge, only three studies have been published to date on HER2-positive metastatic BC (HER2 + mBC), all with contrasting results [8][9][10] , and only one 9 included findings on the newer anti-HER2 drugs pertuzumab and trastuzumab emtansine (T-DM1). In our series, BMI was not associated with outcome, in line with the findings of Martin and colleagues.
However, numerous preclinical studies have suggested that adipose tissue may influence response to anti-HER2 treatment. Adipocytes produce several paracrine and endocrine factors known as adipocytokines, including leptin, adiponectin, tumor necrosis factor (TNF)-α and interleukin 6, potentially affecting tumor growth. Leptin receptor and HER2 are frequently co-expressed in breast cancer cell lines and tumors, interacting with each other and leading to HER2 phosphorylation in response to leptin exposure 21 . Leptin can also transactivate HER2 via epidermal growth factor receptor (EGFR) and Janus-activated kinase 2 activation 22 . Another study 23 hypothesized that leptin sustained HER2 protein levels via an upregulation of the heat shock protein 90 chaperone expression.
Duong et al. 24 observed that pre-adipocytes and adipocytes suppressed trastuzumab-mediated antibodydependent cellular cytotoxicity in breast cancer cells through soluble factor secretion, leading to a protective effect on tumor cells.
Several preclinical studies have evaluated mechanisms of resistance to trastuzumab. One such mechanism is the activation of signaling pathways, in particular the insulin-like growth factor 1 (IGF-1) receptor, which is also known to impede trastuzumab-mediated growth inhibition in breast cancer cells. Saxena et al. 25 described a bidirectional crosstalk between leptin and IGF-1 signaling mediated by transactivation of EGFR promoting BC cell migration and invasion.
Two large meta-analyses of 43 and 82 international clinical studies reported a poorer outcome for very overweight early BC patients with respect to non-obese patients regardless of menopausal or hormone receptor   26,27 . The impact of BMI may differ according to the chemotherapy regimen, as shown by a recent retrospective analysis of a Breast International Group adjuvant trial 28 demonstrating a differential response to docetaxel, but not to other chemotherapy regimens, based on BMI. As mentioned above, BMI alone may not be the most suitable method to assess body composition. Recently, interest has arisen in the role of low muscle mass in cancer patients 29,30 . Individuals with low muscle mass appear to have poorer survival than those with normal muscle mass, especially in gastrointestinal cancer [31][32][33] .
Muscle mass is generally determined by CT scans, which are considered the gold standard for measuring muscle parameters. It is known that the muscle cross-sectional area is strongly correlated with total body muscle mass, and muscle measurement can easily be performed using CT images acquired during routine follow-up. Although the prognostic impact of skeletal muscle measures has been studied in different tumor types, this is relatively unexplored area in BC. The research field of body composition is especially clinically relevant because muscle parameters might help to estimate prognosis and possibly to predict responsiveness to treatments.
In Caan et al. 's study, the largest to date on patients with early BC, muscle and fat mass, evaluated clinically by CT scan, were found to be more strongly correlated with survival than BMI. The authors point out that sarcopenia and adiposity are both important risk factors and should be appraised together when assessing risk of relapse. The HEAL study also reported that sarcopenia, evaluated by DEXA, was linked to overall mortality in 471 patients with stage I-IIIA BC 34 . An impact of muscle mass on chemotherapy toxicity has been reported in a prospective trial on early BC patients treated with taxanes and anthracyclines regimens 35 . www.nature.com/scientificreports/ The majority of data on metastatic disease come from smaller studies focusing on the impact of sarcopenia in patients undergoing chemotherapy 36 . In a study of 55 patients treated with capecitabine for metastatic BC 37 , those who had low muscle mass (calculated using total abdominal muscle area measurement) had a threefold higher risk of toxicity, e.g. diarrhea and stomatitis, and a shorter time to progression. Rier et al. reported that sarcopenia does not portend a poorer prognosis in patients treated with anthracycline and taxane-based chemotherapy regimens. Our findings confirm same results: sarcopenia was not associated with shorter PFS 38 .
To our knowledge, this is the first study that analyzes the association between body composition parameters and response to dual anti-HER2 antibody blockade regimens for metastatic HER2-positive breast cancer. We observed a shorter PFS in patients with increased amounts of total abdominal fat tissue, while muscle mass did not affect outcome.
Our findings differ from those highlighted by Franzoi et al. in patients receiving endocrine therapy plus cyclin-dependent kinase (CDK) 4/6 inhibitors for hormone receptor-positive, HER2-negative metastastic BC: their results showed an association between sarcopenia and a worse PFS, while ours didn't. In our series, total visceral fat is the most important factor related to worse outcome, and it might suggest that in HER2-positive breast cancer disease total abdominal fat tissue could have a key role to predict response to newly anti-HER2 treatments, as already shown in preclinical models 39 .
Dual anti-HER2 antibody blockade with trastuzumab and pertuzumab in association with a taxane is now recommended as first-line systemic treatment in patients with HER2-positive metastatic BC following findings a survival benefit in the Cleopatra Study 11 . A recent update of this trial revealed an association between progesterone receptor expression and outcome. We found a similar trend for progestin receptor and a clearer impact of ER on outcome. The impact of BMI or body composition parameters was not analyzed in the Cleopatra trial 12 .
Our findings provide additional information regarding total abdominal visceral fat tissue as a potential poor prognostic factor in HER2-positive metastatic BC, within the context of dual anti-HER2 antibody blockade treatment. Given that body fat and muscle mass represent modifiable elements, programs that aim to promote or improve exercise and nutritional assessment before therapy should be offered to BC patients.
Our study is exploratory, with limitations due to its retrospective nature and limited sample size. Furthermore, we were unable to recover data on diet, malnutrition, physical activity, sedentary habits and comorbidities, which can influence body composition parameters and their changes over time. Further studies are warranted to better understand the impact of body composition parameters, and their changes over time, on treatment outcomes in the different BC subtypes.  www.nature.com/scientificreports/ Of 63 patients found, 3 received trastuzumab plus pertuzumab in second or later lines, and 2 continued treatment elsewhere, and were excluded from the analysis. Other 15 patients were excluded because baseline imaging was not available (n = 10) or inadequate (n = 5). The final cohort was made up of 43 patients (Fig. 3).

Materials and methods
Body composition parameters assessment. All patients underwent a baseline CT or positron emission tomography (PET)-CT scan as close as possible to the beginning of treatment with anti-HER2 agents (not more than 9 weeks before and 1 week after the start of treatment).
To assess body mass composition, a single CT slice at the level of the third lumbar vertebral body, possibly with both transverse processes depicted, was processed using a specific DICOM-viewer software (OsiriX© v.11.0.0; Pixmeo, Geneve, Switzerland). To standardize the measurements of the BCp, all the CT slice analyzed had these characteristics: acquired at 100-120 kV with variable mA, soft tissue reconstruction algorithm, matrix of 512 × 512 and reconstructed slice thickness of 5 mm. On this slice, using a built-in application enabling a semiautomated tissue demarcation based on density thresholds, the total cross-sectional area of fat tissues (TAFT) was calculated by summing the areas of visceral fat tissue (VF), considering a density threshold from − 150 to − 50 Hounsfield Units and excluding visceral organs, and subcutaneous and intramuscular fat (SF), using a density threshold from − 190 to − 30 Hounsfield Units. Similarly, in the same slice, considering a density threshold from − 29 to + 150 Hounsfield Units, the total cross-sectional area of skeletal muscle (TMA) was evaluated (including paraspinal, psoas, and abdominal wall musculature and excluding bones).
All the measurements were performed by a single reader (API), who manually corrected tissue segmentations when necessary, unaware of the final PFS of patients, to avoid any bias.
To confirm these measurements and verify inter-observer variability, another reader (DB), blinded to the BCp obtained by the first reader, independently performed all the measurements and the agreement between the two readers was assessed.
Patient heights were used to calculate the Skeletal Muscle Index (SMI) as follows 40 : Similarly, sub-cutaneous fat, visceral fat and total abdominal fat were corrected for patient's height and indexes were obtained (SFI, VFI, and TAFTI).
Metabolic and clinical parameters. From patient's medical records, we retrieved several metabolic and clinical baseline parameters such as age, height, weight, BMI, low density lipoprotein (LDL) cholesterol and triglyceride serum levels, menopausal status, ECOG performance status. Pathology data, including histological type, disease stage, Ki67, ER and progestin receptor (PgR) expression levels, were also obtained.
Definition of parameters' categories. While sarcopenia for BC patients is well defined in the literature (SMI < 40) 41 , the remaining body composition parameters (SFI, VFI, and TAFTI) were categorized calculating the median values of the whole cohort and classified in two categories: low (below median value) and high (equal or above the median value). Metabolic and Clinical Parameters were dichotomized using well established thresholds values 42 .
Statistical analysis. Progression-free Survival (PFS) was defined as the time elapsed between the beginning of treatment with anti-HER2 agents and radiologically confirmed disease progression according to RECIST v1.1 or death from any causes, and was tested using Kaplan-Meier estimates. Survival curves were compared