The role of body mass index at diagnosis in patients with inflammatory breast cancer

Abstract

Introduction

Inflammatory Breast Cancer (IBC) is an aggressive presentation of BC present in 1–5 % of all patients with BC. While obesity has been consistently associated with worse prognosis in patients with BC, it is understudied in patients with IBC.

Patients and methods

We retrospectively evaluated the association of body mass index (BMI) at diagnosis with clinicopathological characteristics, pathological complete response (pCR) to chemotherapy and survival in a multicentric cohort of patients with IBC treated with pre-operative chemotherapy.

Results

Of the 542 patients, 6 were underweight (1.1 %, excluded in further analysis), 190 were normal-weight (35.1 %), 187 had overweight (34.5 %), and 159 had obesity (29.3 %). Of the 536 included patients, 463 had non-metastatic and 73 metastatic IBC at diagnosis. Higher BMI was associated with older age at diagnosis, increased stromal tumor infiltrating lymphocytes (sTIL), and particularly in the ER-/HER2+ subgroup, a greater likelihood of metastasis at diagnosis. Tumor emboli were less frequently detected in peritumoral samples of patients with obesity as compared to patients with normal weight. Among non-metastatic patients, those with obesity in the ER-/HER2+ and ER+/HER2-subgroups showed numerically, but not statistically significantly, lower rates of pCR following neoadjuvant chemotherapy than normal-weight patients (38.5 % vs 42.5 %, and 6.6 % vs 16.7 %, respectively). BMI was not associated with any survival endpoint.

Conclusion

We observed a limited association of BMI with clinicopathological variables and pCR in patients with IBC without evidence of its relationship with survival outcomes. Future studies should investigate the biological impact of obesity on IBC and its tumor microenvironment.

Highlights

• •Multicentric retrospective study of 542 patients with inflammatory breast cancer (IBC).
• •Older age is associated with a higher body mass index (BMI) at diagnosis.
• •Patients with higher BMI had metastatic disease more often at diagnosis in the ER−/HER2+ subgroup.
• •A trend toward higher sTIL in patients with higher BMI was seen.
• •No association is observed between BMI and pathological complete response (pCR) rate or survival.

1. Introduction

Inflammatory breast cancer (IBC) is a relatively rare clinical presentation of breast cancer (BC), with prevalence of 1–5 % [[1], [2], [3]]. Despite its rarity, it is an aggressive disease, causing up to 7 % of BC mortality [[1], [2], [3]]. It is mostly a clinical diagnosis characterized by distinct skin changes, such as diffuse erythema and edema (peau d'orange) in at least one-third of the breast, and an abrupt onset and rapid progression of symptoms [4]. Dermal-lymphatic invasion with tumor emboli is seen in 75 % of patients. These emboli are thought to contribute to the clinical presentation and the rapid metastatic spread [4,5]. Axillary lymph node involvement is often present [1]. Up to 25 % of patients with IBC present with primary metastatic disease, compared to only 6 % for patients with non-IBC [6]. Patients with IBC are generally younger and have poorer survival rates compared to those with non-IBC locally advanced BC [7,8]. There is also a higher proportion of the more aggressive molecular subtypes in IBC as compared to non-IBC, with estrogen receptor (ER) negative/human epidermal growth factor receptor 2 (HER2) negative tumors reported in 22–29 % and HER2-positive tumors in 32–45 % of patients with IBC as opposed to 10–15 % and 13–15 % in patients with non-IBC [1,9,10].

Treatment options for IBC are expanding across all molecular subtypes and stages of BC but progress for this aggressive disease has been hampered by the low prevalence and limited number of dedicated studies [11]. The standard treatment for stage III, and sometimes oligometastatic IBC, involves a multimodality approach, including pre-operative systematic therapy, mastectomy, axillary lymph node dissection, and radiotherapy [12]. Five-year overall survival (OS) rates ranges from 44 % in patients with triple-negative IBC to 74 % in patients with HER2+ IBC. The pathological complete response (pCR) rate varies significantly by subtype, influencing survival outcomes with worse outcomes in case of incomplete response [9]. As described in our earlier retrospective study in patients with IBC treated with preoperative systemic therapy, high tumor grade, ER negativity, HER2 positivity, higher stromal tumor infiltrating lymphocytes (sTIL), and taxane-based neoadjuvant chemotherapy (NACT) were significantly associated with pCR and better outcomes were observed in case of pCR [13].

Obesity is an acknowledged risk factor for developing BC, especially in postmenopausal women [[14], [15], [16], [17], [18]]. Regarding IBC, obesity has been recognized as a risk factor, with a higher proportion of patients with obesity at diagnosis compared to patients with non-IBC, particularly in postmenopausal women, but also in pre- and perimenopausal patients [10,19,20]. Obesity is not only associated with an increased risk of developing BC but has also been associated with worse survival outcomes across all molecular subtypes [[21], [22], [23]]. Knowledge about IBC in this regard remains limited due to the scarcity of studies involving patients with IBC or the small number of patients included in studies and the contradictory findings. For instance, Chang et al. reported lower survival if higher body mass index (BMI) but only in postmenopausal women, while Kogawa et al. did not see an effect of BMI on survival [24,25].

The results from the pooled analyses of patients treated with NACT by Fontanella et al. suggested that a higher BMI was associated with lower pCR rates. In addition, they observed that patients with obesity experienced more often non-hematological adverse events, had lower compliance to chemotherapy and received lower doses of taxanes [26]. This is in contrast to the study of Kogawa et al. in IBC where a higher BMI was associated with a higher pCR rate [25]. Higher levels of sTIL have been consistently associated with higher levels of pCR in patients treated with NACT across all molecular subtypes [27]. We have recently demonstrated that BMI modifies the effect of sTIL on pCR and prognosis in patients with triple-negative BC (TNBC) treated with NACT, with high sTIL levels being only associated with pCR in patients with normal weight but not in patients with overweight or obesity [28]. While these studies give us insight into the association between obesity and treatment response, so far it has been understudied whether this would also be the case in patients with IBC.

This study aims at retrospectively evaluating the association of BMI at diagnosis with the clinicopathological characteristics of the tumor, response to neoadjuvant therapy and survival in the largest multicentric cohort of patients with IBC.

2. Methods

2.1. Study design

As previously described [13], this retrospective, multicentric study included a cohort of female patients diagnosed with IBC between October 1996 and October 2021 at eight different European hospitals. Initial patient selection was based on the reported cT4d T-stage [29]. To discriminate between “real” IBC and secondary IBC (IBC-like presentation due to progression of a neglected breast), clinical files were refined for the criteria defined by Dawood and colleagues [30]. Patients were eligible irrespective of the presence of metastatic disease at diagnosis, provided they had received pre-operative chemotherapy. BMI was categorized according to the WHO expert committee guidelines into underweight (<18.5 kg/m²), normal-weight (≥18.5 and < 25 kg/m²), overweight (≥25 and < 30 kg/m²), and obesity (≥30 kg/m²) [31]. Details on data collection, histopathological characterization, and definition of treatment response surrogates, including pCR and residual cancer burden (RCB), are available in the supplementary methods.

2.2. Statistical analysis

Statistical analyses were performed using R version 4.1.1. All statistical tests were two-sided and considered statistically significant when the p-value was <0.05.

The standard clinicopathological variables considered in this study were age (>50 vs. ≤50 years), tumor grade (G3 vs. G1/G2), nodal involvement (yes vs. no), metastatic disease at diagnosis (yes vs. no), ER expression (positive vs. negative), and HER2 status (positive vs. negative). For analyses related to treatment response and survival, treatment variables included the NACT regimen (taxane-based vs. non-taxane) and the use of neoadjuvant anti-HER2 therapy (yes vs. no).

The associations of BMI as a categorical dependent variable with standard clinicopathological variables and sTIL were assessed using multinomial regression models with ‘normal-weight’ as the reference category. Analyses involving BMI as a continuous dependent variable were performed using linear regression models in a similar manner. The association of pCR, RCB class, and RCB score with BMI was evaluated strictly in patients without metastatic disease at diagnosis using Firth's logistic regression, multinomial regression, and linear regression models, respectively. Two models were applied in each regression analysis: Model 1, adjusted for center, and Model 2, which additionally accounted for standard clinicopathological variables and, when analyzing pCR and RCB, treatment variables.

Survival analyses were performed in patients without metastatic disease at diagnosis. The median follow-up duration was estimated using the reverse Kaplan–Meier method. Three key survival endpoints were defined: disease-free survival (DFS), the interval from diagnosis to the first event after diagnosis of either locoregional recurrence, contralateral recurrence, distant recurrence, or death from any cause; distant recurrence-free survival (DRFS), the interval from diagnosis to the first event of distant recurrence; and overall survival (OS), the interval from diagnosis to death from any cause. Data on primary non-breast malignancies and their related survival events were unavailable and therefore could not be considered in our analyses. The Kaplan–Meier method was first used to estimate the rates of DFS and OS across different BMI categories. Crude cumulative incidence curves accounting for death without distant recurrence as the sole competing event were constructed for inspection of event rates of DRFS according to BMI categories. Cox regression models were next performed to quantify the association of BMI either as a continuous or categorical variable with DFS and OS: Model 1 was stratified by center, and Model 2 was adjusted for standard clinicopathological and treatment variables and stratified by center. DRFS was analyzed in the presence of death without distant recurrence as the competing risk using Fine-Grey sub-distribution hazard regression models: Model 1 was adjusted center, and Model 2 was additionally adjusted for standard clinicopathological variables, treatment variables, and pCR.

3. Results

3.1. Clinicopathological characteristics of the patient cohort

Data on BMI at diagnosis were available for 542 patients diagnosed with IBC. Among these patients, 6 were classified as having underweight (1.1 %), 190 with normal weight (35.1 %), 187 with overweight (34.5 %), and 159 with obesity (29.3 %). Underweight patients were not included in further analysis due to their low number, leaving a total of 536 patients for the study. Among them, 189 (38.0 %) were diagnosed with ER+/HER2− IBC, 120 (24.1 %) with ER-/HER2- IBC and 189 with HER2+ IBC, corresponding to 76 ER + HER2+ (15.3 % of all patients) and 113 ER- HER2+ (22.7 % of all patients, Table 1).

Table 1.

Clinicopathological characteristics of patients in the entire study cohort and in each BMI category.

		All (N = 536)	Normal weight (N = 190)	Overweight (N = 187)	Obesity (N = 159)	p-value
		n (%)	n (%)	n (%)	n (%)
Age	≤50	213 (39.7)	92 (48.4)	66 (35.3)	55 (34.6)	0.014
	>50	323 (60.3)	98 (51.6)	121 (64.7)	104 (65.4)
Menopausal status	Pre/Peri-menopausal	217 (43.6)	90 (50.8)	67 (39.0)	60 (40.3)	0.047
	Post-menopausal	281 (56.4)	87 (49.2)	105 (61.0)	89 (59.7)
	Unknown	38	13	15	10
Histology	ILC	32 (6.7)	17 (10.0)	9 (5.5)	6 (4.3)	0.309
	IBC-NST	433 (91.2)	149 (87.6)	152 (92.1)	132 (94.3)
	Other	10 (2.1)	4 (2.4)	4 (2.4)	2 (1.4)
	Unknown	61	20	22	19
Grade	1	17 (3.5)	8 (4.7)	4 (2.4)	5 (3.4)	0.266
	2	153 (31.7)	48 (27.9)	50 (30.1)	55 (37.9)
	3	313 (64.8)	116 (67.4)	112 (67.5)	85 (58.6)
	Unknown	53	18	21	14
ER status	Negative	242 (47.0)	87 (48.1)	90 (49.7)	65 (42.5)	0.387
	Positive	273 (53.0)	94 (51.9)	91 (50.3)	88 (57.5)
	Unknown	21	9	6	6
PR status	Negative	314 (63.2)	109 (62.6)	114 (64.8)	91 (61.9)	0.865
	Positive	183 (36.8)	65 (37.4)	62 (35.2)	56 (38.1)
	Unknown	39	16	11	12
HER2 status	Negative	318 (61.2)	111 (62.0)	107 (59.8)	100 (64.5)	0.679
	Positive	195 (38.0)	68 (38.0)	72 (40.2)	55 (35.5)
	Unknown	23	11	8	4
Surrogate subtype	ER-/HER2-	120 (24.1)	41 (23.6)	43 (24.9)	36 (23.8)	0.823
	ER-/HER2+	113 (22.7)	43 (24.7)	42 (24.3)	28 (18.5)
	ER+/HER2-	189 (37.9)	66 (37.9)	62 (35.8)	61 (40.4)
	ER+/HER2+	76 (15.3)	24 (13.8)	26 (15.0)	26 (17.2)
	Unknown	38	16	14	8
Nodal involvement	No	90 (17.3)	34 (18.6)	26 (14.3)	30 (19.5)	0.399
	Yes	429 (82.7)	149 (81.4)	156 (85.7)	124 (80.5)
	Unknown	17	7	5	5
Metastasis at diagnosis	No (M0)	463 (86.4)	165 (86.8)	159 (85.0)	139 (87.4)	0.813
	Yes (M1)	73 (13.6)	25 (13.2)	28 (15.0)	20 (12.6)
Pre-operative chemotherapy scheme	No Taxane	101 (19.0)	46 (24.6)	28 (15.0)	27 (17.1)	0.056
	Taxane	431 (81.0)	141 (75.4)	159 (85.0)	131 (82.9)
	Unknown	4	3	0	1
Pre-operative anti-HER2	No	399 (75.3)	142 (76.3)	137 (73.3)	120 (76.4)	0.731
	Yes	131 (24.7)	44 (23.7)	50 (26.7)	37 (23.6)
	Unknown	6	4	0	2
Surgery	Mastectomy	370 (96.9)	121 (98.4)	131 (96.3)	118 (95.9)	0.570
	Tumorectomy	12 (3.1)	2 (1.6)	5 (3.7)	5 (4.1)
	Unknown	154	67	51	36
pCR	Yes	137 (26.9)	48 (27.1)	52 (29.4)	37 (23.9)	0.541
	No	372 (73.1)	129 (72.9)	125 (70.6)	118 (76.1)
	Unknown	27	13	10	4
RCB class#	0	42 (27.6)	16 (31.4)	15 (26.3)	11 (25.0)	0.253
	I	13 (8.5)	2 (3.9)	8 (14.0)	3 (6.8)
	II	50 (32.9)	20 (39.2)	19 (33.3)	11 (25.0)
	III	47 (30.9)	13 (25.5)	15 (26.3)	19 (43.2)
	Unknown	21	8	10	3

# Statistics are described for patients diagnosed and treated at University Hospitals Leuven.

Abbreviations: BMI, body mass index; ER, estrogen receptor; HER2, Human Epidermal growth factor Receptor 2; IBC-NST, invasive breast cancer of no special type; ILC, invasive lobular carcinoma; pCR, pathological complete response; PR, progesterone receptor; RCB, residual cancer burden.

The most common histological subtype was invasive breast cancer of no special type (IBC-NST, 433 patients, 91.2 %), followed by invasive lobular carcinoma (ILC, 32 patients, 6.7 %) (Table 1). Locoregional lymph node involvement at time of diagnosis was seen in most patients (429 patients, 83 %, Table 1). At diagnosis, 463 patients (86 %) presented without metastases, while 73 patients (14 %) had metastatic disease at primary diagnosis (M1, Table 1). Among patients with available treatment data, the majority received taxane-based pre-operative chemotherapy (431/532 patients, 80 %), mastectomy (370/382 patients, 97 %), and radiotherapy (434/455, 95 %). Differences in the treatment scheme between the different BMI categories were not observed (Table 1). Within the HER2+ groups, 36.0 % (68/189) did not receive pre-operative anti-HER2 therapy because the date of diagnosis preceded the introduction of anti-HER2 as a standard BC treatment (Table 1). pCR was observed in 27 % of patients with available data (137/509) and was the highest in the ER-/HER2+ subgroup (46.6 %, Table 1).

Center-specific differences were observed in several patient characteristics, including BMI (Supplementary Table 1). Consequently, all subsequent regression analyses were adjusted for center.

A central pathology review was conducted on 400 unique samples from 370 patients (69.0 %). There were no significant differences in clinicopathologic variables or BMI between patients with and without a central pathology review (Supplementary Table 1). sTIL were scored in 368 samples from 350 of patients undergoing central pathology review (94.6 %). For patients with multiple biopsies, the mean sTIL score across all samples was used as the representative value at the patient level [13]. The median sTIL was 5.3 % (IQR: 2.0 %–16.7 %), with the majority of cases (n = 242, 64.2 %) classified as having low sTIL, defined by a 10 % cutoff (Supplementary Fig. 1). Patients in the ER−/HER2− group had the highest sTIL (median: 10.0 %; IQR: 5.0 %–20.0 %), while those in the ER+/HER2− group had the lowest (median: 3.3 %; IQR: 0.7 %–10.3 %; Supplementary Fig. 1A–B). Tumor emboli were identified in 71 of 338 patients (21.0 %), and 36 of 60 patients (60.0 %) with evaluable peritumoral samples and skin biopsies, respectively (Supplementary Fig. 1C).

3.2. Association of BMI at primary diagnosis with clinicopathological characteristics of IBC

We first examined the association between BMI, either as a categorical or continuous variable, and clinicopathological features of IBC. In the overall cohort, BMI was significantly associated with age, with patients over 50 years more likely to have a higher BMI and be classified into the overweight or obesity categories compared to those aged 50 years or younger (Table 1, Fig. 1). This association was consistently observed in both patients with M0 and M1 disease (Fig. 1). Among patients with M0 disease but not those with M1, a slightly higher proportion of those with obesity had ER + disease compared to normal-weight patients (59.4 % vs. 50.6 %, Supplementary Table 2), without reaching statistical significance (Fig. 1). Among patients with M1 disease, those with a lower BMI tended to have high-grade tumors (normal-weight: 86.1 %, overweight: 69.6 %, obesity: 50 %, Supplementary Table 3, Fig. 1). No clear associations were found between other standard clinicopathological features and BMI (Fig. 1, Supplementary Tables 2–3).

Fig. 1.

Association of BMI with clinicopathological features in all patients. (A–B) Forest plots showing the association of BMI (categorical) (A) and BMI (continuous) (B) with standard clinicopathological variables in the overall cohort and the M0 and M1 sub-cohorts. The odds ratios shown in (A) are log-scaled.

Abbreviations: BMI, body mass index; ER, estrogen receptor; HER2, Human Epidermal growth factor Receptor 2; M0, no metastasis at diagnosis; M1, metastasis at diagnosis.

Stratified analyses by clinical subtype based on ER and HER2 expression further revealed an association between higher BMI and a greater likelihood of metastatic disease at diagnosis in the ER-/HER2+ (normal-weight: 11.6 %, overweight: 16.7 %, obesity: 28.6 %, Supplementary Figs. 2–3). No notable associations were additionally observed in the other subgroups.

When analyzing the relationship between BMI and sTIL levels, we observed a general trend where higher BMI was associated with increased sTIL scores (Fig. 2). However, when considering sTIL as a categorical variable, this association was primarily observed between patients with obesity and normal weight (Fig. 2D). Specifically, tumors from patients with obesity were more likely to have intermediate/high sTIL levels (42.9 %) compared to those from normal-weight patients (30.6 %, Fig. 2B–D). This finding appeared to be driven by the M0 sub-cohort, which represented the majority of the study population, as a similar trend was not observed in the M1 sub-cohort. Further analyses revealed that this positive association between BMI and sTIL was only present in the ER+/HER2-and ER-/HER2+ subgroups of the overall cohort and the M0 sub-cohort (Supplementary Figs. 4–5). The association was more pronounced in the ER+/HER2-subgroup, while weaker statistical evidence was seen in the ER-/HER2+ subgroup. Additionally, these associations were more evident when sTIL was analyzed as a continuous variable rather than a categorical one.

Fig. 2.

sTIL scoring in different BMI categories and its association with BMI in all patients. (A–B) Distribution of sTIL (continuous) (A) and sTIL (categorical) (B) across BMI categories in the overall cohort. (C–D) Forest plots showing the association of BMI (categorical) with sTIL (continuous) (C) and sTIL (categorical) (D) in the overall cohort and the M0 and M1 sub-cohorts. The odds ratios shown in (D) are log-scaled. (E–F) Forest plots showing the association of BMI (continuous) with sTIL (continuous) (E) and sTIL (categorical) (F) in the overall cohort and the M0 and M1 sub-cohorts.

Abbreviations: BMI, body mass index; M0, no metastasis at diagnosis; M1, metastasis at diagnosis; sTIL, stromal tumor infiltrating lymphocytes.

In contrast to sTIL, tumor emboli were less frequently detected in peritumoral samples from patients with higher BMI, with a more pronounced difference between patients with obesity and normal weight than between patients with overweight and normal weight (Supplementary Fig. 6A–D). This association remained perceptible in the ER-/HER2-and ER+/HER2+ subgroups when considering clinical subtypes (Supplementary Fig. 7). In cases with detected tumor emboli in the overall cohort and M0 sub-cohort, BMI was negatively associated with emboli density; however, this was not confirmed in the regression model adjusted for other clinicopathological features (Supplementary Fig. 6E–F). Due to the limited number of cases, analyses for the M1 sub-cohort were not considered feasible. Regarding tumor emboli evaluated in skin biopsies, no significant differences were observed in the overall or M0 groups.

3.3. Association of BMI at primary diagnosis with response to NACT and survival of IBC

Given the challenges in accurately assessing treatment response at metastatic sites and the unknown status of such assessment in the current cohort, we focused our investigation of the potential association between BMI and response to NACT, as well as survival outcomes, on patients without metastatic disease at diagnosis (M0). pCR rates were 26.9 % in patients with normal weight, 30.8 % in patients with overweight, and 23.9 % in patients with obesity. Regression analyses showed no statistically evident association between BMI and either pCR or RCB (Fig. 3). However, subgroup analyses suggested a negative association between BMI and pCR in patients with ER-/HER2+ disease when BMI was analyzed as a continuous variable, though this lacked sufficient statistical evidence at the conventional significance level (Supplementary Fig. 8). When BMI was categorized, patients with obesity in the ER-/HER2+ and ER+/HER2-subgroups had lower rates of achieving pCR compared to their normal-weight counterparts (38.5 % vs 42.5 %, and 6.6 % vs 16.7 %, respectively). However, these associations were not statistically confirmed by regression models. A possible interaction between BMI and sTIL on pCR was explored by introducing an interaction term into the regression models, but did not present statistical significance (p-value = 0.867 for all evaluable cases). Of note, the sub-cohort available for this analysis (with complete data for sTIL scoring and all covariates), comprising only 277 cases, was not fully representative of the study cohort. In this subset, while the BMI distribution and pCR rates were similar to the complete M0 cohort, normal-weight patients were slightly more likely to achieve pCR (29.0 % vs. 26.9 %), while patients with obesity were less likely (21.7 % vs. 23.9 %).

Fig. 3.

Association of pCR and RCB with BMI in all patients. (A–B) Forest plots showing the association of pCR with BMI (categorical) (A) and BMI (continuous) (B) in the M0 sub-cohort. The odds ratios shown are log-scaled. (C–D) Forest plots showing the association of RCB class with BMI (categorical) (C) and BMI (continuous) (D) in the M0 sub-cohort. The odds ratios shown are log-scaled. (E–F) Forest plots showing the association of RCB score with BMI (categorical) (E) and BMI (continuous) (F) in patients with residual disease (RCB score >0) in the M0 sub-cohort.

Abbreviations: BMI, body mass index; M0, no metastasis at diagnosis; , pCR, pathological complete response; RCB, residual cancer burden.

The median follow-up time for patients with M0 disease was 9.35 years. Kaplan-Meier curves for DFS and OS revealed no significant differences in prognosis between patients with overweight or obesity and patients with normal weight (Fig. 4A and B). Furthermore, both univariable and multivariable regression models did not reveal any prognostic significance of BMI (Fig. 4C and D). Similarly, DRFS analysis did not show any BMI-associated differences (Supplementary Fig. 9).

Fig. 4.

Association of DFS and OS with BMI in patients with M0 disease. (A) Kaplan-Meier curves of DFS according to BMI category in the M0 sub-cohort. (B–C) Forest plots showing the association of DFS with BMI (category) (B) and BMI (continuous) (C) in the M0 sub-cohort. The hazard ratios shown are log-scaled. (D) Kaplan-Meier curves of OS according to BMI category in the M0 sub-cohort. (E–F) Forest plots showing the association of OS with BMI (category) (E) and BMI (continuous) (F) in the M0 sub-cohort. The hazard ratios shown are log-scaled.

Abbreviations: BMI, body mass index; M0, no metastasis at diagnosis; DFS, disease-free survival; OS, overall survival.

4. Discussion

In this study, we present the largest retrospective analysis focusing on BMI in patients with IBC treated with pre-operative chemotherapy.

Associations between clinicopathological variables and BMI at diagnosis were limited to age, with higher age at diagnosis being associated with a higher BMI at diagnosis. This is comparable to findings in retrospective, non-IBC focused studies [32,33]. This could be expected as a link between older age and higher BMI is seen in the general population as well [34]. We described a greater likelihood of metastatic disease at diagnosis in the ER−/HER2+ subgroup if higher BMI in IBC. A negative prognostic effect of obesity in the ER-/HER2+ subgroup has already been described in early BC, but contradictory findings are seen in advanced BC [35].

In studies across breast cancer stages, patients with obesity have worse outcomes with lower rates of breast-cancer-specific survival and OS than patients without obesity [21,22]. We did not see this in our cohort nor in any analyzed subgroups. This is as opposed to the study of Chang et al. including 177 patients with IBC where shorter overall survival if higher BMI in IBC was seen but only among postmenopausal women [24]. This study included patients between 1974 and 1993 and although these patients received NACT, therapies have strongly evolved since then. Our results are in line with those of Kogawa et al. with patient inclusion between 2006 and 2012 [25].

Obesity and IBC have both been linked to worse outcomes and resistance to therapy through impaired immune response and a more pro-inflammatory state [36,37]. For example, in both, lower production of the cytokine IL-10 was seen which is linked to BC growth and tumor invasion [36,38]. Given these overlapping biological features, any additive effect of BMI on outcomes may be relatively modest and difficult to detect within this already aggressive disease.

A pCR rate of 27 % was observed across the cohort, which is slightly higher than described in other IBC studies (11–23 %) [[39], [40], [41]]. Although most patients were diagnosed before 2010 when neoadjuvant anti-HER2 treatment became standard of care, the majority with HER2+ disease still received it, contributing to higher pCR rates in HER2+ subgroups compared to HER2-ones. In the literature, patients overweight and obesity with operable BC had lower pCR rates with NACT compared to patients with under-/normal weight [42]. However, a small study by Kogawa et al. reported the opposite trend in patients with IBC [25]. In our study, no statistically significant association was seen between pCR and BMI. However, patients with obesity in specific subgroups, including ER-/HER2+ and ER+/HER2-, had numerically lower rates of achieving pCR compared to their normal-weight counterparts. These observations did not reach statistical significance, possibly because of the relatively low number of patients. One possible contributor to lower pCR rates in patients with obesity may be suboptimal chemotherapy exposure due to capped body surface area-based dosing, a practice employed by some but not all participating centers. However, due to missing dosing data for most patients in our cohort, this potential confounder could not be fully assessed but partially accounted for through adjustment by center. These warrant further investigation in larger, well-documented cohorts.

In our study, we observed a trend between higher BMI and increased sTIL scores, mostly when comparing patients with obesity to patients with normal weight. Noteworthy, the association between higher sTIL and BMI was only preserved in specific subgroups, including ER-/HER2+ where paradoxically, patients with obesity showed lower pCR rates. In the general BC population, higher sTIL scores are associated with increased pCR rates in patients treated with NACT and with a favorable prognosis in patients with TNBC and HER2+ BC [27,43]. However, in patients with obesity, it has been observed that the effect of sTIL is modified with a loss of effect on a favorable prognosis in TNBC [28]. Further research on obesity-driven immune modulation in IBC may shed light on the biology underlying these observations.

While higher sTIL are seen in patients with obesity, less tumor emboli in peritumoral samples were observed in these patients. Several hypotheses could explain this observation. First, as most reviewed slides were obtained from biopsies rather than resection specimens, in adipose-rich tissue, the likelihood of encountering emboli through a biopsy might be reduced. Second, previous studies have shown that emboli are more commonly found in lymphatic vessels than in blood vessels [44], and lymphatic vessel density tends to decrease with obesity [45]. These however remain hypotheses, as our dataset did not allow us to evaluate them.

This study presents with several limitations. First, this is an early cohort, with patient inclusion beginning in 1996. Therefore, data on more recent treatments were not available, limiting our insights on responses and outcomes related to these treatments. Specifically, almost all patients were treated before immunotherapy was available in the neoadjuvant setting for TNBC and 35 % of patients with HER2+ disease did not receive anti-HER2 treatment as part of their neoadjuvant regimen while both are now standard of care [46]. Furthermore, patients with locally advanced hormone receptor positive disease are sometimes treated with neoadjuvant endocrine therapy (NET) combined with CDK4/6 inhibitors, while patients receiving NET were not included in this study. In the future, the prognosis of patients with IBC could potentially be improved by therapies that enhance the pCR rate and precisely target distinct features of the IBC TME. Second, since this was not a nested case-control study, no comparison could be made with the non-IBC patients from the different institutions. Third, we only captured and used BMI at diagnosis, while BMI is a dynamic variable that can change during life and after diagnosis. Furthermore, BMI is not considered to be fully representative of the effect of adiposity on patients and their disease [47]. Lastly, although studies have shown an effect of race and ethnicity on the incidence and outcomes of IBC [48], we could not address these potential confounders in our analyses due to their unavailability in the dataset.

Nevertheless, this study relies on two major strengths. It represents the largest retrospective cohort including multiple centers with central pathology review performed for most samples. Additionally, patients were screened for ‘true’ IBC based on the Dawood criteria which were not always available or used in earlier retrospective studies [30].

Prospective trials are needed to validate these findings. These trials should include markers for adiposity other than BMI. Future research is also needed to understand how adiposity could impact the composition and phenotype of the different cells from the tumor microenvironment, as well as their interactions. For instance, we reported differential enrichment of contradictory immune-related pathways in various immune and stromal cell types in the primary breast tumor microenvironment, suggesting multidirectional inflammation associated with obesity [49]. This could advance our understanding of whether there would be a potential need and benefit to tailor treatment of patients with IBC according to their adiposity.

5. Conclusion

In contrast to findings in the general BC population, our study did not identify statistically significant association between obesity and pCR following NACT or survival outcomes in patients with IBC. However, the observed link between BMI and sTIL suggests that obesity may play a biologically relevant role in IBC, which could be different from that in non-IBC. Future research is needed to further explore the interplay between obesity and the TME to uncover potential clinical implications in this aggressive disease.

CRediT authorship contribution statement

Kristien Borremans: Writing – review & editing, Writing – original draft, Methodology, Investigation. Ha-Linh Nguyen: Writing – review & editing, Writing – original draft, Visualization, Software, Methodology, Investigation, Formal analysis, Data curation. Maxim De Schepper: Writing – review & editing, Writing – original draft, Investigation, Funding acquisition, Data curation. Florence Lerebours: Writing – review & editing, Resources, Investigation. Roman Vion: Writing – review & editing, Resources, Investigation. Florian Clatot: Writing – review & editing, Resources, Investigation. Anca Berghian: Writing – review & editing, Resources, Investigation. Marion Maetens: Writing – review & editing, Resources, Project administration, Methodology, Funding acquisition. Edoardo Isnaldi: Writing – review & editing, Resources, Investigation. Chiara Molinelli: Writing – review & editing, Resources, Investigation. Matteo Lambertini: Writing – review & editing, Resources, Investigation. Federica Grillo: Writing – review & editing, Resources, Investigation. Gabriele Zoppoli: Writing – review & editing, Resources, Investigation. Luc Dirix: Writing – review & editing, Resources, Investigation. Kevin Punie: Writing – review & editing, Resources, Investigation. Hans Wildiers: Writing – review & editing, Resources, Investigation. Ann Smeets: Writing – review & editing, Resources, Investigation. Ines Nevelsteen: Writing – review & editing, Resources, Investigation. Patrick Neven: Writing – review & editing, Resources, Investigation. Anne Vincent-Salomon: Writing – review & editing, Resources, Investigation. Denis Larsimont: Writing – review & editing, Resources, Investigation. Caroline Duhem: Writing – review & editing, Resources, Investigation. Patrice Viens: Writing – review & editing, Resources, Investigation. François Bertucci: Writing – review & editing, Resources, Investigation. Elia Biganzoli: Writing – review & editing, Methodology. Peter Vermeulen: Writing – review & editing, Resources, Investigation. Giuseppe Floris: Writing – review & editing, Supervision, Resources, Methodology, Conceptualization. François Richard: Writing – review & editing, Supervision, Software, Formal analysis. Christine Desmedt: Writing – review & editing, Visualization, Supervision, Project administration, Methodology, Funding acquisition, Conceptualization.

Ethical approval

The study was approved by the ethics committee of University Hospital Leuven on December 20, 2019 (S62499) and conducted in accordance with the Declaration of Helsinki. No informed consent form was requested as a waiver was granted given that many patients already passed away or progressed.

Data availability statement

The data generated in this study are not publicly available due to restrictions of the protocol approved by the ethics committees of the involved hospitals. Access to the data can be requested via the corresponding author.

Funding sources

This study received financial support by the Fondation Cancer Luxembourg (FC/2018/07) and Fonds Wetenschappelijk Onderzoek – Vlaanderen (FWO, G059821N). Additionally, K.B. and M.D.S. were funded by the fund Nadine de Beauffort, H-L.N. and M.M. and by the European Research Council, M.D.S. and M.M. by the Luxemburg Cancer Foundation, and M.D.S., H.W., G.F. and F.R. were funded by the FWO. The funders played no role in study design, data collection, analysis and interpretation of data, or the writing of this manuscript.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Christine Desmedt reports financial support was provided by Fondation Cancer Luxembourg. Christine Desmedt reports financial support was provided by Research Foundation Flanders. Kristien Borremans, Maxim De Schepper reports a relationship with Fund Nadine de Beauffort that includes: funding grants. Ha-Linh Nguyen, Marion Maetens reports a relationship with European Research Council that includes: funding grants. Maxim De Schepper, Marion Maetens reports a relationship with Luxemburg Cancer Foundation that includes: funding grants. Maxim De Schepper, Hans Wildiers, Giuseppe Floris, Francois Richard reports a relationship with Research Foundation Flanders that includes: funding grants. Florian Clatot reports a relationship with AstraZeneca, Daiichi Sankyo, Gilead, MSD, Merck Serono, Nutricia, Novartis that includes: consulting or advisory and travel reimbursement. that includes: consulting or advisory and travel reimbursement. Chiara Molinelli reports a relationship with Daiichi Sankyo, Seagen that includes: consulting or advisory. Chiara Molinelli reports a relationship with Accademia Nazionale di Medicina that includes: speaking and lecture fees. Chiara Molinelli reports a relationship with Menarini that includes: travel reimbursement. Matteo Lambertini reports a relationship with Roche, Lilly, Novartis, Astrazeneca, Pfizer, Seagen, Gilead, MSD, Exact Sciences, Pierre Fabre, Menarini that includes: consulting or advisory. Matteo Lambertini reports a relationship with Roche, Lilly, Novartis, Pfizer, Sandoz, Libbs, Daiichi Sankyo, Takeda, Menarini, AstraZeneca that includes: speaking and lecture fees. Matteo Lambertini reports a relationship with Gilead, Daiichi Sankyo, Roche that includes: travel reimbursement. Matteo Lambertini reports a relationship with Gilead that includes: funding grants. Kevin Punie reports a relationship with Astra Zeneca, Eli Lilly, Exact Sciences, Focus Patient, Gilead Sciences, Medscape, MSD, Mundi Pharma, Need Inc., Novartis, Pfizer, Hoffmann-La Roche, Sanofi, Seagen that includes: consulting or advisory and speaking and lecture fees. Kevin Punie reports a relationship with Gilead Sciences, MSD that includes: travel reimbursement. Hans Wildiers reports a relationship with Daiichi Sankyo, Gilead, Lilly, Pfizer, Novartis, PSI, Augustine Therapeutics, Astra Zeneca, Roche, Agendia, Immutep, Seagen that includes: consulting or advisory. Hans Wildiers reports a relationship with Daiichi Sankyo that includes: travel reimbursement. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is/are the supplementary data to this article.