Quantitative dynamic contrast-enhanced MRI parameters effectively predict treatment efficacy of neoadjuvant chemotherapy in breast cancer

Abstract

Purpose: To investigate the predictive value of quantitative DCE-MRI parameters for estimating the treatment efficacy of neoadjuvant chemotherapy (NACT) in breast carcinoma (BC). Methods: A retrospective analysis was conducted on 178 pathologically confirmed cases of BC, diagnosed via puncture biopsy, at The Second Affiliated Hospital of Anhui Medical University between January 2019 and June 2023. All patients received preoperative NACT. Based on postoperative pathological inspection results, 53 patients with grade IV-V pathological responses were included in the major histological response (MHR) group, and the remaining 125 with grade I-III pathological responses were assigned to the non-major histological response (NMHR) group. The pre- and post-chemotherapy early-phase enhancement rate (E₁), peak enhancement rate (E_max), and time to peak (T_max) on DCE-MRI were compared between the two patient cohorts. Quantitative parameters such as volume transfer constant (K^trans), rate constant (K_ep) and extravascular extracellular volume fraction (V_e) were obtained, and the post-NACT maximum tumor diameter (D-max) reduction rate and tumor volume reduction rate (TVRR) were calculated. Furthermore, the predictive efficacy of pre- and post-NACT quantitative DCE-MRI parameters for treatment responses was evaluated using receiver operating characteristic (ROC) curves. Results: The MHR group showed statistically higher post-NACT D-max reduction rate and TVRR than the NMHR group. The two patient cohorts were similar in pre-chemotherapy K_ep, but the pre-chemotherapy K^trans and V_e were lower in MHR; the post-chemotherapy K^trans, K_ep and V_e were all statistically different between groups (P < 0.05). The MHR group presented markedly lower E₁ and E_max values and statistically longer T_max compared to the NMHR group after NACT (all P < 0.05). The pre-NACT quantitative DCE-MRI parameters demonstrated limited prediction performance, with V_e showing the highest efficacy (AUC = 0.612); in contrast, post-NACT quantitative DCE-MRI parameters exhibited improved predictive accuracy, with K^trans demonstrating the best predictive performance (AUC = 0.801). Conclusions: The pre-NACT quantitative DCE-MRI parameters are not effective in predicting the therapeutic outcome of NACT. However, the post-NACT DCE-MRI parameters provide accurate and reliable predictions of pathological responses, with K^trans showing the highest predictive value and considerable clinical applicability.

Introduction

Breast carcinoma (BC) is the leading cause of morbidity and mortality among women, with an increasing incidence in younger age groups. This common malignant tumor significantly impacts women’s physical and mental health and poses a serious threat to their lives [1]. In China, BC is among the malignancies with the fastest-growing fatality rate [2]. Early prevention and treatment are crucial for improving patient outcomes, with the current standard of care relying on a combination of surgical intervention, radiotherapy and chemotherapy, targeted therapy, endocrine therapy, and immunotherapy [3-6]. However, many patients are diagnosed at intermediate or advanced stages, resulting in poor prognoses due to the lack of specific symptoms in early-stage BC.

The advent of neoadjuvant chemotherapy (NACT) has revolutionized the treatment for locally advanced BC [7]. Preoperative NACT can effectively reduce tumor size and downgrade the tumor stage, increasing the possibility of breast-conserving surgery and the likelihood of achieving a postoperative pathological complete response (pCR) [8]. NACT has demonstrated high clinical response rate (80%), with pCR observed in 6-25% of BC patients [9,10]. A 2014 meta-analysis of 11,955 cases published in The Lancet highlighted that patients achieving pCR after NACT experienced improved event-free and overall survival, with pCR being strongly associated with long-term survival benefits [11]. Thus, accurate and timely evaluation of the pathological response to NACT is crucial for guiding treatment decisions. Histopathological examination remains the “gold standard” evaluating tumor response [12], offering high diagnostic accuracy. However, it can only be performed post-surgery, which may result in missing the optimal window for adjusting treatment plans. Thus, the ability to dynamically assess tumor responses to NACT in vivo is vital, enabling timely adjustments for both responders and non-responders.

Various imaging modalities, including ultrasound, mammography, and breast magnetic resonance imaging (MRI), have been applied to evaluate the pathological response of BC patients to NACT [13-15]. While numerous studies have sought to determine the optimal diagnostic model for evaluating NACT efficacy, no consensus has been reached [16-18]. Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is a minimally invasive and widely applied imaging technique [19], enabling quantitative evaluation of the hemodynamic changes in tumors. It reflects the biological response of the tumor during treatment and provides a more accurate and earlier assessment of the tumor’s response compared to purely morphological evaluations. This technique can assist clinical evaluations and predict the efficacy of chemotherapeutic drugs. Meanwhile, DCE-MRI has significant advantages, as it not only reflects changes in microvessel density and tumor hemodynamics during chemotherapy, but also offers detailed information on tumor’s angiogenic status. By directly introducing contrast agents or using pharmacokinetic models for perfusion analysis, three parameters, namely volume transfer constant (K^trans), rate constant (K_ep) and extravascular extracellular volume fraction (V_e), can be calculated to provide insights into tumor blood perfusion and microvascular permeability [20]. K^trans reflects local blood volume, blood flow, and vascular permeability; K_ep correlates with local blood flow and vascular permeability; and V_e is associated with cell density and microvessel density. It has been reported that K^trans and V_e are closely related to microvessel density [21]. Previous studies have generally indicated that quantitative DCE-MRI parameters after two cycles of chemotherapy can predict patient outcomes [22-24]. However, controversies still exist regarding the effectiveness of DCE-MRI in predicting NACT efficacy for BC. In this study, DCE-MRI was performed on BC patients before and after NACT to assess the predictive value of its parameters and their correlation with post-NACT pathological responses.

Materials and methods

Study population

This retrospective study included 178 BC patients, all of whom were pathologically confirmed by core needle biopsy at The Second Affiliated Hospital of Anhui Medical University and received preoperative NACT. Inclusion criteria: (1) Diagnosis of breast cancer by core needle biopsy before NACT, and confirmed by pathology biopsy after surgery; (2) No prior BC surgery or systemic chemotherapy before the first MRI examination; (3) Clinical staging of IIA-IIIC, with distant metastasis excluded by imaging; (4) No previous history of cancer; (5) Receipt of three or more cycles of NAC after diagnosis, followed by curative-intent surgery, such as breast-conserving surgery or modified radical mastectomy; (6) Completion of both pre- and post-NACT DCE-MRI of the breasts; (7) Availability of complete medical records. Exclusion criteria: (1) Imaging artifacts severe enough to affect evaluation; (2) Previous cancer history; (3) Chemotherapy intolerance or switch to an alternative chemotherapy regimen during treatment; (4) Absence of postoperative pathological examination results due to lack of post-chemotherapy surgical treatment; (5) Incomplete medical records.

The NACT scheme primarily consisted of anthracyclines or taxanes, including the TEC regimen (paclitaxel, epirubicin, and cyclophosphamide), the PC regimen (paclitaxel and carboplatin), and the FAC scheme (5-fluorouracil, anthracycline, and cyclophosphamide). The choice of NACT regimen for patients was based on tumor characteristics, patient’s physical condition, allergic history, and the physician’s experience with the medications. This study was approved by the Ethics Committee of The Second Affiliated Hospital of Anhui Medical University.

Inspection methods

All patients underwent DCE-MRI both before and after NACT. Prior to the examination, patients were instructed to fast for 6 hours, and were advised against physical exercise or coughing during the procedure. The examination was performed using Siemens 3.0T MRI scanner equipped with an 8-channel phased array breast coil. During the scan, patients were positioned in the prone position with the breasts hanging naturally within the surface coil; intravenous access was established using a 12G needle for the hyperbaric injector before the examination. The scanning procedures were as follows. Plain Scan: T1-weighted imaging (T1WI), adipose-suppressed T2WI in the axial plane, diffusion weighted imaging (DWI) (b = 0, b = 800 s/mm²). Dynamic Contrast-Enhanced Imaging: 3D fast spoiled gradient echo sequence (FLASH-3D) and T1WI fat-suppressed axial imaging, parameters: TR/TE 4.56 ms/1.5 ms, flip angel = 12°, field of view (FOV) = 360 mm, matrix = 384×288, slice thickness = 1 mm. A total of 5 phases were scanned continuously, with each phase collected over 90 seconds, for a total duration of 7 minutes and 30 seconds. The procedure included an initial “mask” scan, followed by contrast injection with gadopentetate dimeglumine (0.2 mL/kg), and a saline flush of 20 mL at a flow rate of 2 mL/s. An immediate dynamic enhancement scan was performed following the injection.

Image analysis

All data were analyzed using Siemens Mean Cure software and Siemens Vida syngo MR XA20 post-processing software. Rhe acquired images were imported into the workstation, and the region of interest (ROI) was selected from the area showing the most significant enhancement and the largest lesion size during continuous contrast-enhanced scanning. To ensure consistency, the same ROI was used before and after chemotherapy. Care was taken to avoid necrotic and vascular areas during ROI selection. The software automatically calculated the following parameters: early-phase enhancement rate (E₁), peak enhancement rate (E_max), time to peak (T_max), volume transfer constant (K^trans, mmol/min), rate constant (K_ep, mmol/min; rate constant of inter-tissue contrast agents returning to the blood vessels), and extravascular extracellular volume fraction (V_e; the volume fraction of the extravascular extracellular space). For each lesion ROI, the software automatically generated quantitative parameter values. Measurements were repeated three times, and the average values were recorded. Care was taken to avoid cystic and necrotic areas. Additionally, the maximum tumor diameter (D-max) and tumor volume were determined, and the D-max reduction rate and tumor volume reduction rate (TVRR) post-NACT were calculated as follows: D-max or TVRR = (Pre-chemotherapy value - Post-chemotherapy value)/Pre-chemotherapy value × 100%.

Pathological evaluation

A pathologist with over 10 years of experience was tasked with reviewing the tissue specimens in a blinded manner. Pathological results from the surgical sections of all patients were graded according to the Miller & Payne grading system. The pathological responses were classified into either non-major histological response (NMHR) or major histological response (MHR). NMHR: grade I - No change in the number of tumor cells before and after NACT; grade II - Tumor cell density decreased by ≤ 30% after NACT; grade III - Tumor cell density decreased by 30%-90% after NACT. MHR: grade IV - Tumor cell density decreased by ≥ 90% after NACT; grade V - Complete disappearance of tumor cells, with no residual invasive carcinoma.

Statistical methods

The primary quantitative parameters for analysis were DCE-MRI parameters, while the secondary parameters were the histological response grades. SPSS 22.0 software and Graphpad Prism 8 were used for statistical analyses. The categorical data were represented by counts (%) and analyzed using the χ² test. The measurement data were expressed as mean ± standard deviation. Normality of |the data was assessed first; if the data followed a normal distribution, comparisons were made using the t-test. If the data were not normally distributed, non-parametric tests were used. The prediction performance was assessed using receiver operating characteristic (ROC) curves and the area under the curve (AUC), which was further compared using DeLong test. The difference was considered statistically significant at p-value < 0.05.

Results

General information of the included patients

A total of 178 BC patients included in this study underwent postoperative pathological evaluation, with 17, 73, 62, 19, and 7 cases of stages IIA, IIB, IIIA, IIIB, and IIIC, respectively. Of these, 53 patients were treated with TEC regimen, 48 patients with PC regimen, and 77 patients with FAC regimen. There were 53 patients in MHR group and 125 patients in NMHR group, as shown in Table 1.

Table 1.

General information of selected patients

Clinicopathological features	Number of cases (n = 178)
Age (years old)	46.9±6.2
Maximum lesion diameter (cm)	3.7±1.1
Lesion distribution
Left breast	87
Right breast	91
Clinical staging
IIA	17
IIB	73
IIIA	62
IIIB	19
IIIC	7
Neoadjuvant chemotherapy regimen
TEC regimen	53
PC regimen	48
FAC regimen	77
Pathological evaluation of chemotherapy
Non-major histological response	125
Major histological response	53

Note: TEC regimen (paclitaxel, epirubicin, and cyclophosphamide); PC regimen (paclitaxel and carboplatin); FAC regimen (5-fluorouracil, anthracycline, and cyclophosphamide).

Comparison of D-max reduction rate and TVRR between the two groups

As shown in Figure 1, MHR patients exhibited notably higher post-NACT D-max reduction rate and TVRR than the NMHR patients (both P < 0.05).

Figure 1.

Comparison of maximum tumor diameter reduction rate and tumor volume reduction rate. A: Maximum tumor diameter reduction rates; B: Tumor volume reduction rates. MHR, Major histological response; NMHR, Non-major histological response; ***P < 0.001.

Pre- and post-NACT quantitative DCE-MRI parameters in the two groups

The comparison of pre- and post-NACT quantitative parameters of DCE-MRI is shown in Table 2. Before chemotherapy, the MHR group showed similar K_ep values (P > 0.05) but lower K^trans and V_e values (P < 0.05) compared to the NMHR group. After chemotherapy, all three indices, K^trans, K_ep and V_e, were significantly lower in the MHR group compared to those in the NMHR group (all P < 0.001). In the MHR group, K^trans, K_ep and V_e values were significantly lower post-chemotherapy compared to their pre-chemotherapy values (all P < 0.05), while in the NMHR group, the V_e values remained unchanged before and after chemotherapy (P > 0.05).

Table 2.

Comparison of pre- and post-NACT quantitative DCE-MRI parameters

	Parameters	NMHR group (n = 125)	MHR group (n = 53)	t/Z	P
Before chemotherapy	Ktrans (min-1)	0.322±0.112	0.280±0.119	2.247	0.026
	Kep (min-1)	0.938±0.256	0.860 (0.683, 1.116)	1.368	0.173
	Ve	0.462±0.101	0.412±0.115	2.913	0.004
	E1 (%)	87.8±13.0	102.2±15.9	6.311	< 0.001
	Emax (%)	138.7±23.6	158.7±23.3	5.189	< 0.001
	Tmax (s)	151.3±17.1	64.5±9.7	34.631	< 0.001
After chemotherapy	Ktrans (min-1)	0.252±0.091*	0.166±0.051*	6.514	< 0.001
	Kep (min-1)	0.633±0.155*	0.510±0.098*	5.320	< 0.001
	Ve	0.439±0.103	0.375±0.086*	3.985	< 0.001
	E1 (%)	101.6±14.5*	62.8±6.9*	18.586	< 0.001
	Emax (%)	165.9±20.5*	138.7±18.9*	8.280	< 0.001
	Tmax (s)	131.8±10.0*	152.7±15.5*	10.721	< 0.001

Note: NACT, Neoadjuvant chemotherapy; DCE-MRI, Dynamic contrast-enhanced magnetic resonance imaging; MHR, Major histological response; NMHR, Non-major histological response; K^trans, Volume transfer constant; K_ep, Rate constant; V_e, Extravascular extracellular volume fraction; E₁, Early-phase enhancement rate; E_max, Peak enhancement rate; T_max, Time to peak;

^*P < 0.05, versus before chemotherapy within the group.

Pre- and post-NACT DCE-MRI signal parameters of patients in the two groups

Before NACT, the MHR group exhibited significantly higher E₁ and E_max values and shorter T_max compared to the NMHR patients (all P < 0.05). After NACT, the MHR group showed marked reductions in E₁ and E_max, but statistically prolonged T_max, compared to their pre-NACT levels (all P < 0.05). In contrast, the NMHR group demonstrated significant increases in E₁ and E_max but notably shorter T_max compared to their pre-NACT levels (all P < 0.05). In addition, the inter-group comparison revealed lower E₁ and E_max while longer T_max in MHR group as compared to NMHR group (all P < 0.001), as shown in Table 2.

A typical case is illustrated in Figure 2. A 48-year-old female patient was diagnosed with right breast cancer after a 3-year history of a right breast mass. MRI showed a mass-like soft tissue signal in the upper quadrant of the right breast, characterized by low signal in the T1WI sequence, slightly high signal in the T2WI sequence, and longer T2 signal in the inner part of the breast. The DWI sequence showed a high signal, and the ADC map showed a low signal, with irregular morphology, lobular shape, poorly defined borders, and an area of about 3.9×5.4×5.2 cm. The mass was adjacent to thickened skin of the right breast, which exhibited uneven enhancement after contrast injection, with a rich blood supply and multiple tortuous thickened blood vessels visible on maximal intensity projection (MIP). After 4 months of neoadjuvant chemotherapy, the lesion significantly reduced in size, and the T1WI, T2WI, and DWI sequences were not clear. Post-chemotherapy enhancement showed flaky enhancement foci.

Figure 2.

MRI images of a 48-year-old female case before and after neoadjuvant chemotherapy. A: Before neoadjuvant chemotherapy; B: 4 months after neoadjuvant chemotherapy. DWI, Diffusion weighted imaging; ADC, Apparent dispersion coefficient; MIP, Maximal intensity projection.

Predictive efficacy of pre- and post-NACT quantitative DCE-MRI parameters

The predictive efficacy of pre-NACT quantitative DCE-MRI parameters in predicting NACT efficacy is shown in Figure 3 and Table 3. The pre-NACT parameters demonstrated poor prediction efficiency, with V_e being the most effective (AUC = 0.612), showing a sensitivity of 56.6% and a specificity of 63.2%. Post-NACT quantitative DCE-MRI parameters showed improved predictive performance, with K^trans providing the greatest predictive efficiency (AUC = 0.801, sensitivity of 83.0%, and specificity of 72.0%). The prediction performance of K_ep was moderate, with an AUC of 0.738, while that of V_e value was the least, with an AUC of 0.689.

Figure 3.

ROC curves for pre- and post-NACT DCE-MRI parameters in predicting major histological responses to NACT. A: Pre-NACT DCE-MRI parameters in predicting major histological responses; B: Post-NACT DCE-MRI parameters in predicting major histological responses. NACT, Neoadjuvant chemotherapy; DCE-MRI, Dynamic contrast-enhanced magnetic resonance imaging; ROC, Receiver operating characteristic.

Table 3.

Predictive efficiency of quantitative DCE-MRI parameters before and after neoadjuvant chemotherapy

	Parameters	AUC	Cut-off	Maximum Youden index	Sensitivity (%)	Specificity (%)	P
Before chemotherapy	Ktrans (min-1)	0.591	0.326	0.191	51.2	67.9	0.056
	Kep (min-1)	0.583	0.911	0.264	52.8	73.6	0.110
	Ve	0.612	0.441	0.198	56.6	63.2	0.018
After chemotherapy	Ktrans (min-1)	0.801	0.204	0.550	72.0	83.0	< 0.001
	Kep (min-1)	0.738	0.643	0.394	48.8	90.6	< 0.001
	Ve	0.689	0.432	0.328	55.2	83.0	< 0.001
Delong test	Before chemotherapy	After chemotherapy	Difference in AUC	SE	95% CI	Z	P
	Ktrans (min-1)	Ktrans (min-1)	0.210	0.059	0.094-0.327	3.540	< 0.001
	Kep (min-1)	Kep (min-1)	0.155	0.059	0.038-0.273	2.597	< 0.001
	Ve	Ve	0.077	0.058	-0.036-0.190	1.338	0.181

Note: DCE-MRI, Dynamic contrast-enhanced magnetic resonance imaging; K^trans, Volume transfer constant; K_ep, Rate constant; V_e, Extravascular extracellular volume fraction; AUC, Area under curve; SE, Standard error.

Discussion

NACT has become the standard treatment for locally advanced BC. As a systemic cytotoxic therapy administered before operation, NACT aims to downstage both the primary tumor and lymph nodes [25], thus increasing the possibility for breast-conserving surgery. A proper evaluation of NACT efficacy not only helps guide the treatment plans but also provides insights into tumor responses to chemotherapy, offering a reliable theoretical basis for postoperative treatment decisions. Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) can provide valuable information about tumor intercellular space, cell composition, blood supply, and vascular permeability, all of which can reflect pathological responses and predict survival outcomes to some extent.

DCE-MRI has been widely used in numerous studies to evaluate and predict the response of breast tumors to NACT [26-28]. Early research primarily focused on semi-quantitative analyses, using enhanced MRI to measure changes in tumor volume [29,30]. Cheung et al. [31] analyzed differences in pre- and post-treatment tumor sizes after a chemotherapy cycle, as well as final tumor size remissions in 33 patients. Their findings revealed that all complete responders experienced a tumor size reduction of more than 45% following one treatment cycle. In this study, 53 patients were classified as MHR and 125 as NMHR. Comparing the tumor shrinkage between the two cohorts revealed that both the D-max reduction rate and TVRR were notably higher in MHR group compared to the NMHR group. The evaluation of treatment efficacy is primarily based on changes in D-max, which is easy to measure, reproducible, and highly accurate. It correlates well with pathological outcomes and is therefore an effective indicator of treatment response and pathological changes in patients undergoing NACT. While these results are promising for BC treatment, morphological changes, such as tumor size and qualitative enhancement patterns, are temporal consequences of underlying physiological alterations. Consequently, changes in tumor perfusion parameters may serve as early surrogate biomarkers for treatment response. Thus, alongside tumor size evaluation, researchers have increasingly focused on the quantitative physiological parameters provided by DCE-MRI [26].

Tumor tissue typically contains a large number of neovascular vessels that are immature and lack endothelial cells or smooth muscle cells, resulting in high vascular permeability. BC patients with better NACT response generally have fewer blood vessels in the tumor and lower permeability, significantly compromising tumor’s blood supply and perfusion [32]. It has been shown that the growth and metastasis of breast cancer depend on a rich blood supply, and that local blood supply and capillary permeability play a crucial role in the absorption of chemotherapeutic drugs [33,34]. Changes in K^trans and k_ep can therefore reflect the tumor’s response to treatment efficacy.

Quantitative parameters such as K^trans, K_ep, and V_e were compared in this study, and the post-NACT reductions of K^trans, K_ep, and V_e were more prominent in MHR patients compared to NMHR patients. During chemotherapy, tumor cell necrosis can reduce vascular endothelial growth factor production, leading to apoptosis of immature vascular endothelial cells and reduced neovascularization, which results in a significant decrease in K^trans and K_ep [35]. Interestingly, we found no statistical difference between pre- and post-chemotherapy V_e in the NMHR group. This may be due to the instability of V_e measurement, as the surrounding edema of the lesion has a strong influence. Thus, the use of V_e in efficacy evaluation remains controversial [36]. Besides, while pre-chemotherapy K_ep was similar between the two cohorts, K^trans and V_e were lower in the MHR group as compared to the NMHR group, which may be related to measurement errors in K^trans, as factors affecting blood perfusion, like hypertension, can affect K^trans value, leading to inaccuracies [37]. As there is a relationship between V_e and K^trans and K_ep (V_e = K^trans/K_ep), any errors in K^trans measurement can propagate to V_e calculations.

Furthermore, the pre-NACT E₁ and E_max in MHR patients were noticeably higher compared with NMHR patients, while the pre-NACT T_max was markedly shorter. After NACT, MHR patients showed statistically lower E₁ and E_max, along with a longer T_max than NMHR patients. In patients with effective NACT, changes in tumor microvessels and permeability often result in residual tumor lesions being largely unenhanced, causing hemodynamic parameters to resemble those of normal tissue. This leads to a significant reduction in E₁ and E_max, as well as an extension of T_max. Moreover, ROC curve analysis revealed that the predictive efficiency of pre-NACT quantitative DCE-MRI parameters was poor, with V_e being the most effective (AUC = 0.612). However, post-NACT, the predictive performance of these parameters improved, with K^trans showing the best predictive efficacy (AUC = 0.801). The discrepancies observed in existing studies may be attributed to differences in scanning machines, temporal resolutions, scanning protocols, chemotherapy regimens, and experimental grouping criteria.

Still, there are some limitations that need to be addressed. First, this study is a single-center retrospective analysis, with a small sample size and relatively narrow pathological scope, which limits its generalizability. Second, some patients had non-mass-like lesions, making it difficult to depict ROI in multi-phase axial dynamic enhanced images. Although the values of all parameters were averaged, there are still potential deviations from the actual values of lesions. Furthermore, the number of MHR cases was smaller than the number of NMHR cases, which could influence the analysis of group differences. Despite these limitations, this study provides valuable information for the clinical assessment of the prognosis of neoadjuvant chemotherapy in BC patients. Moreover, given that DCE-MRI is a noninvasive test, its application holds higher potential for clinical convenience and value. Conclusively, quantitative DCE-MRI parameters are effective in predicting pathological responses following NACT for BC, providing valuable prognostic information for BC patients undergoing NACT.

Disclosure of conflict of interest

None.