The Role of Magnetic Resonance Imaging in Predicting Pathological Complete Response in Breast Cancer Patients Receiving Neoadjuvant Chemotherapy

Abstract

Objective

Neoadjuvant chemotherapy (NACT) has been used more frequently in the treatment of non-metastatic breast cancer. Achieving a pathological complete response (pCR) after NACT provides us with information about the prognosis of the disease. We aimed to investigate the diagnostic accuracy of preoperative magnetic resonance imaging (MRI) in predicting pCR in patients with breast cancer.

Method

The charts of patients who received NACT for breast cancer between 2018 and 2024 were retrospectively analyzed. Factors associated with radiologic complete response (rCR) and pCR were analyzed in univariate and multivariate settings. The correlation between rCR and pCR according to MRI interpretation was analyzed.

Results

A total of 153 patients were included in this study; 26 (17%) of these patients achieved rCR, and 39 (25.5%) achieved pCR. A statistically significant correlation was found between pCR and rCR assessed on MRI. Sensitivity, specificity, accuracy, pCR predictive value (PPV) and non-pCR predictive value were estimated at 51.3%, 94.7%, 83.7%, 76.9% and 85%, respectively. Statistically significant correlations between rCR and pCR were found in Luminal A (p<0.05), Luminal B (p<0.05), HER2-positive (p<0.05), but not in triple negative subtypes (p>0.05). In univariate and multivariate analysis, tumor characteristics significantly associated with both rCR and pCR were tumor size, lymph node metastasis and molecular subtypes.

Conclusion

MRI evaluation varies according to the subtype and after NACT showed a correlation between rCR and pCR.

Introduction

Breast cancer remains the most commonly diagnosed type of cancer worldwide (1). Traditional treatment methods (surgery + adjuvant chemotherapy) have an important place in the treatment approach, but neoadjuvant chemotherapy (NACT) is increasingly discussed and implemented in current breast cancer treatment guidelines.

NACT refers to systemic chemotherapy administered prior to planned surgery for patients with locally advanced breast cancer. Its primary goal is to transform initially inoperable tumors into operable ones, thereby offering patients the chance for surgical intervention and improving breast conservation rates. Additionally, NACT enables early assessment of the chemotherapy’s impact on cancer tissue, which helps guide the selection of the most effective chemotherapy agents for the patient’s future treatment (2).

When the data, in the literature, are examined, it has been observed that high pathological complete response (pCR) rates are achieved, when effective targeted therapies and chemotherapeutic agents are preferred in first-line treatment (3, 4).

Currently, no standardized methods or imaging biomarkers are available in clinical practice to accurately predict pCR after NACT. Magnetic resonance imaging (MRI) is considered the most sensitive imaging modality that can be used to assess treatment response in patients undergoing NACT (5).

Radiomics is a feature extraction and analysis method commonly used in medical imaging. Radiomics analyzes pixel values in medical images and inter-pixel relationships.

It is thought that the application of radiomics techniques to MRI imaging will contribute positively to all stages of breast cancer with the additional data it will provide to the diagnosis, treatment and prognosis processes of breast cancer (6).

Scientific studies suggest that patients with radiologic complete response (rCR) show significant correlation with pCR (7).

Nevertheless, there is a limited number of studies investigating the relationship between rCR and pCR across various molecular subtypes, particularly where the Ki-67% proliferation index is used as a criterion.

In this study, we aimed to evaluate the effectiveness of post-treatment breast MRI in forecasting pCR following NACT using a large dataset from a single center.

Material and Methods

Patient Characteristics

We retrospectively analyzed breast MRI data of breast cancer patients with four or more cycles of NACT, at University of Health Sciences Turkey, İstanbul Training and Research Hospital between March 2018 and May 2024. The baseline data for our study were collected from the hospital’s medical records database. We received approval from the University of Health Sciences Turkey, İstanbul Training and Research Hospital’s Clinical Research Ethics Committee (date: 19.07.2024, number: 36).

The study included patients with histopathologically confirmed, radiologically non-metastatic breast cancer who had received NACT. Patients who underwent MRI to monitor tumor response to NACT and had at least two MRI evaluations was incorporated into the study. The initial MRI was conducted following the diagnosis and prior to NACT, and the second MRI was performed after completion of NACT, before surgery.

Patients who could not complete the NACT process had incomplete follow-up data were excluded. In Table 1, the baseline characteristics of the patients are displayed.

Most patients in the study received a combination of anthracycline and cyclophosphamide followed by taxane chemotherapy.

Along with these chemotherapeutic agents, HER2-positive breast cancer patients were also treated with targeted therapy using pertuzumab and trastuzumab. Some patients with triple-negative breast cancer received a carboplatin regimen combined with paclitaxel chemotherapy.

Pathological Classification

According to receptor positivity, tumors were categorized as estrogen receptor (ER) positive or progesterone receptor (PR) positive if the immunohistochemical (IHC) staining of the tumor tissue showed 1% or more positivity in tumor cell nuclei. The HER2 status (positive or negative) was assessed through IHC and/or fluorescence in situ hybridization (FISH) analysis of biopsy samples. In the IHC analysis, a score of 3+ was deemed positive, while scores of 0 to 1+ were considered negative. For tumors that received a 2+ score, further evaluation via FISH was conducted to check for HER2 gene amplification.

The Ki-67% index was divided into three categories based on values: Less than 15%, between 15% and 40%, and more than 40%. Tumors were classified into molecular subtypes using IHC markers. The subtypes included Luminal A (ER-positive, PR ≥20%, HER2-negative, Ki-67% ≤15%), Luminal B (ER-positive, PR <20% or both ER and PR-positive, HER2-negative, or ER/PR/HER2-positive, with Ki-67% >15%), HER2-positive (ER and PR-negative, HER2-positive), and triple-negative (ER/PR/HER2-negative) groups. The Luminal B group was further divided into two subgroups based on the Ki-67% index: Those with Ki-67% <40% were classified as Luminal B low Ki-67%, while those with Ki-67% ≥40% were labeled as Luminal B high Ki-67%.

In postoperative pathology, the Miller-Payne grading system was utilized to assess the response to NACT. Grades 1 and 2 were classified as stable response; grades 3 and 4 as partial pathological response; and grade 5 as pCR. Residual ductal carcinoma in situ was excluded from the definition of pCR, irrespective of lymph node metastasis presence.

Breast MRI and Evaluation

MRIs were obtained using a 1.5 T MAGNETOM Aera device (Siemens, GERMANY). According to the breast MRI protocol, axial anterior contrast T1 with 3 mm slice thickness, axial STIR T2, sagittal T2 with 4 mm slice thickness for both breasts, and ADC maps and DWI b values (50-400-800 s/mm²) were obtained. Dynamic sequences were acquired six times at 60-second intervals after contrast agent injection, and subtraction images were obtained from these sequences. Finally, a contrast fat-suppressed sagittal T1 VIBE sequence was obtained. MRI examinations were performed in a prone position using a breast coil. To minimize background contrast, patients were imaged between the tenth and fourteenth day of their menstrual cycle.

Statistical Analysis

SPSS 28.0 software system was used for the analyses, and a p-value of less than p<0.05 was considered statistically significant. Statistics of the data included mean, maximum, minimum, median, ratio values, frequency, and standard deviation. The Shapiro-Wilk and Kolmogorov-Smirnov tests were used for assessing the distribution of variables. The Mann-Whitney U test was used to analyze independent quantitative data with abnormal distribution. The chi-square test was used for analyzing qualitative independent data, and if the chi-square test was not satisfied, Fischer’s test was applied. Specificity, sensitivity and accuracy were calculated to evaluate the effectiveness of MRI after NACT. Multivariable regression was used to examine the simultaneous effects of multiple factors. The accuracy of MRI was measured using predictive values, including negative predictive value (NPV) and positive predictive value (PPV).

Results

The files of 570 patients who underwent NACT for breast cancer at the Oncology Clinic of University of Health Sciences Turkey, İstanbul Training and Research Hospital were analyzed. One hundred fifty-three patients with available MRI images and postoperative pathology results were included. Table 1 shows the baseline characteristics of the patients. The average age of the patients was 57.4±12.9 years, with 88.9% diagnosed with invasive ductal carcinoma. Among them, 22 patients (14.4%) were HER2-positive, 24 patients (15.7%) were triple-negative, and 107 patients (69.9%) were hormone receptor-positive.

Following NACT, all patients underwent surgery. The tumor diameter measured on pretreatment MRI was 4.26±1.95 cm (ranging from 1.5 to 10.6 cm), while the residual tumor diameter measured on post-NACT MRI was 2.41±1.31 cm (ranging from 0 to 6.5 cm).

Out of the cohort, 39 patients achieved pCR, and 26 patients had rCR, a statistically significant finding (p<0.05). The sensitivity, accuracy (ACC), specificity, PPV for pCR, and NPV were estimated to be 51.3%, 83.7%, 94.7%, 76.9%, and 85%, respectively (Table 2).

Additionally, we explored the correlation between rCR and pCR across different breast cancer subtypes. Statistically significant correlations between pCR and rCR were observed in the Luminal A, Luminal B, and HER2-positive groups, whereas no significant correlation was found in the triple-negative group.

Discussion

NACT plays a crucial role in managing locally advanced breast cancer. One study has clearly demonstrated that pCR achieved with NACT improves survival and is a key endpoint for improved prognosis (8). However, the definition of pCR continues to be debated, and there seems to be a lack of consensus in the literature, which complicates the objective comparison of results. In our study, we considered the lack of DCIS or malignant cells in the sections taken from the tumor site as pCR. In our study, we found that rCR rates and pCR rates were similar. We aimed to evaluate whether rCR had a significant relationship with pCR. We also wanted to determine the factors that may affect rCR and pCR.

From Table 3, we determined that tumor size measured by MRI after NACT was the most significant factor associated with both pCR and rCR. Other factors, including primary tumor size by MRI, nodal metastasis, and molecular subtype, were also significantly linked to pCR and rCR. This correlation may be attributed to the fact that smaller tumor size, earlier clinical classification, and higher proliferation rates are associated with a greater likelihood of response and a higher chance of achieving pCR and rCR post-NAC (9).

We also aimed to evaluate the precision of MRI in predicting pCR after NACT, particularly for each breast cancer subtype. We based our classifications on IHC markers used in both literature and clinical practice (10). Additionally, we categorized Luminal B breast cancer into two groups based on Ki-67% scores: Ki-67 >20% and Ki-67 <20%. Our findings indicated that rCR was significantly correlated with pCR, especially in the Luminal B, low and high Ki-67%, HER2-positive, and Luminal A subtypes. Nonetheless, we observed notable differences among these three groups.

Moreover, Luminal B, high Ki-67% and HER2-positive subtypes exhibited low NPV, but high PPV. This implies that residual lesions identified by MRI serve as reliable markers of non-pCR for these subtypes; however, the rCR diagnosed by MRI in these groups may be overestimated. Conversely, MRI predicted pCR in the triple-negative subtype with lower sensitivity, specificity, accuracy, PPV, and NPV, compared to the other subtypes, which is contrary to the findings reported in existing literature (11, 12).

In our study, Luminal A and Luminal B breast cancers with low Ki-67% (<20%) were less likely to exhibit pCR or rCR following NACT. However, they achieved high specificity, sensitivity, ACC, NPV and PPV. These data were similar to the results of studies in the literature (13). Certainly, multicenter studies with a larger cohort of patients are needed to speak more precisely. However, MRI still demonstrated a good NPV for these two molecular subtypes. It indicates that the accuracy remains high if the MRI reveals a residual tumor. Some studies have shown that rCR assessed on MRI is not a clear marker for pCR (14, 15). The biggest difference between our data and the data in these studies was that we applied the IHC classification methods recommended in the guidelines. We also set a more objective standard for evaluating rCR. rCR was described as the lack of DCIS and malignant cells. This may have contributed to the disparity between our study and the previously mentioned studies.

While pCR remains the gold standard for assessing pathology specimens after NACT, rCR can serve as a predictor of improved prognosis in clinical practice (16). However, there are challenges associated with diagnosing rCR following NACT. The treatment can induce reactive changes such as inflammation and fibrosis, which may result in non-specific contrast enhancement, leading to either underestimation or overestimation of tumor diameter on MRI (17).

Moreover, different molecular subtypes of breast cancer exhibit varying capabilities in achieving pCR. Ki-67% is a key marker for gauging response to neoadjuvant treatment, when Ki-67% exceeds 40%, the rate of pCR after NACT significantly increases, along with a higher rate of rCR diagnoses. In the Luminal B subtype, a Ki-67% ≥40% indicates a higher likelihood of achieving pCR, with rates comparable to those observed in HER2-positive breast cancer. Although the differences in molecular subtype did not significantly impact the diagnosis of rCR, high Ki-67% emerged as a strong predictor of rCR when compared to low Ki-67%.

Study Limitations

Our study has a number of limitations. The limitations are, firstly, a retrospective design and, secondly, a small sample size. As this was a single-center study, we aimed to employ consistent data analysis methods for both radiological and pathological evaluations. Third, our study did not include immunotherapy agents the guidelines.

Conclusion

In conclusion, the precision of breast MRI in estimating pCR was high in patients receiving NACT for breast cancer. NPV, sensitivity, PPV and precision of MRI in predicting pCR differed significantly between molecular subtypes of breast cancer.

Ethics

Ethics Committee Approval: We received approval from the University of Health Sciences Turkey, İstanbul Training and Research Hospital’s Clinical Research Ethics Committee (date: 19.07.2024, number: 36).

Informed Consent: Retrospective study.

Authorship Contributions

Surgical and Medical Practices: İ.G., E.Ö., E.S.A., Ö.H., Concept: İ.G., B.G., M.B.U., Ç.U., R.U.G., Design: İ.G., B.G., E.S.A., Ç.U., R.U.G., Data Collection or Processing: İ.G., E.Ö., M.B.U., Analysis or Interpretation: İ.G., E.S.A., Ç.U., R.U.G., Literature Search: İ.G., E.Ö., M.B.U., Writing: İ.G., B.G., Ö.H.

Conflict of Interest: No conflict of interest was declared by the authors.

Financial Disclosure: The authors declared that this study received no financial support.

References

Ferlay J, Colombet M, Soerjomataram I, Mathers C, Parkin DM, Pi&ntilde;eros M, et al. Estimating the global cancer incidence and mortality in 2018: GLOBOCAN sources and methods. Int J Cancer. 2019;144(8):1941-1953.

Conforti F, Pala L, Sala I, Oriecuia C, De Pas T, Specchia C, et al. Evaluation of pathological complete response as surrogate endpoint in neoadjuvant randomised clinical trials of early-stage breast cancer: systematic review and meta-analysis. BMJ. 2021;375:e066381.

Gianni L, Pienkowski T, Im YH, Roman L, Tseng LM, et al. Efficacy and safety of neoadjuvant pertuzumab andtrastuzumab in women with locally advanced, inflammatory, or early HER2-positive breast cancer (NeoSphere): a randomised multicentre, open-label, phase 2 trial. Lancet Oncol. 2012;13(1):25-32.

Wang H, Mao X. Evaluation of the efficacy of neoadjuvant chemotherapy for breast cancer. Drug Des Devel Ther. 2020;14:2423-2433.

Expert Panel on Breast Imaging; Slanetz PJ, Moy L, Baron P, diFlorio RM, Green ED, et al. ACR appropriateness criteria&reg; monitoring response to neoadjuvant systemic therapy for breast cancer. J Am Coll Radiol. 2017;14(11S):S462-S475.

Yu Y, He Z, Ouyang J, Tan Y, Chen Y, Gu Y, et al. Magnetic resonance imaging radiomics predicts preoperative axillary lymph node metastasis to support surgical decisions and is associated with tumor microenvironment in invasive breast cancer: A machine learning, multicenter study. EBioMedicine. 2021;69:103460.

Williams M, Eatrides J, Kim J, Talwar H, Esposito N, Szabunio M, et al. Comparison of breast magnetic resonance imaging clinical tumor size with pathologic tumor size in patients status post-neoadjuvant chemotherapy. Am J Surg. 2013;206:567-573.

Cortazar P, Zhang L, Untch M, Mehta K, Costantino JP, Wolmark N, et al. Pathological complete response and long-term clinical benefit in breast cancer: the CTNeoBC pooled analysis. Lancet. 2014;384:164-172.

Giuliano AE, Connolly JL, Edge SB, Mittendorf EA, Rugo HS, Solin LJ, et al. Breast cancer-major changes in the American Joint Committee on Cancer eighth edition cancer staging manual. CA Cancer J Clin. 2017;67:290-303.

Goldhirsch A, Winer EP, Coates AS, Gelber RD, Piccart-Gebhart M, Th&uuml;rlimann B, et al. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol. 2013;24(9):2206-2223.

Chen JH, Bahri S, Mehta RS, Carpenter PM, McLaren CE, Chen WP, et al. Impact of factors affecting the residual tumor size diagnosed by MRI following neoadjuvant chemotherapy in comparison to pathology. J Surg Oncol. 2014;109(2):158-167.

Yu N, Leung VWY, Meterissian S. MRI performance in detecting pCR after neoadjuvant chemotherapy by molecular subtype of breast cancer. World J Surg 2019;43(9):2254-2261.

Dave RV, Millican-Slater R, Dodwell D, Horgan K, Sharma N. Neoadjuvant chemotherapy with MRI monitoring for breast cancer. Br J Surg. 2017;104(9):1177-1178.

Sener SF, Sargent RE, Lee C, Manchandia T, Le-Tran V, Olimpiadi Y, et al. MRI does not predict pathologic complete response after neoadjuvant chemotherapy for breast cancer. J Surg Oncol. 2019;120(6):903-910.

Weber JJ, Jochelson MS, Eaton A, Zabor EC, Barrio AV, Gemignani ML, et al. MRI and prediction of pathologic complete response in the breast and axilla after neoadjuvant chemotherapy for breast cancer. J Am Coll Surg. 2017;225(6):740-746.

Gampenrieder SP, Peer A, Weismann C, Meissnitzer M, Rinnerthaler G, Webhofer J, et al. Radiologic complete response (rCR) in contrast-enhanced magnetic resonance imaging (CE-MRI) after neoadjuvant chemotherapy for early breast cancer predicts recurrence-free survival but not pathologic complete response (pCR). Breast Cancer Res. 2019;21(1):19.

Woodhams R, Kakita S, Hata H, Iwabuchi K, Kuranami M, Gautam S, et al. Identification of residual breast carcinoma following neoadjuvant chemotherapy: diffusion-weighted imaging--comparison with contrast-enhanced MR imaging and pathologic findings. Radiology. 2010;254(2):357-366.