Synopsis

This study evaluated the use of pre-chemo MRI of axillary lymph nodes (aLNs) to detect metastasis in breast cancer with PET and pathology as reference standards. PET versus pathology reports agreed 87% for presence of disease in the aLNs, while PET and MRI reports agreed 93%. We found good agreement between MRI and PET scores (Cohen’s Kappa 𝜅= 0.39, p=10-5). The MRI sensitivity, specificity, NPV and accuracy were 66.1%, 72.1%, 73.1% and 69.9%, respectively. Although improvement is needed, MRI staging of aLN is possible and practical. This approach may prove helpful in diagnosis and treatment planning of breast cancer patients.

Introduction

In pre-neoadjuvant chemotherapy (NAC) staging of breast cancer patients, ultrasound-guided or MRI-guided needle biopsy of the axillary lymph nodes (aLN) is critically important [1]. Its limitations are that only a small amount of tissue from one or two aLNs can be sampled, potentially missing cancer, and the majority of the biopsied nodes are negative [2], questioning its routine use.

A few FDG-PET studies have shown promising detection accuracy of aLN metastasis [3,4]. However, FDG-PET requires injection of a radiotracer and is not a common for breast cancer patients. A few studies have reported MRI staging of aLN metastasis but the accuracy is moderate (81.6%) [4, 5]. A challenge is a lack of reference standard for validation. While pathology validation is possible, it is difficult in practice to match the nodes on biopsy and MRI because nodes are small and clustered together. Thus, the potential of accurate MRI staging of aLN remains elusive.

The goal of this study was to test the hypothesis that pre-NAC MRI staging of aLNs can be used to accurately detect aLN metastasis associated with breast cancer, by using PET and pathology as reference standards.

Methods

This study included 50 NAC breast cancer patients from 2011-2017, of which 45 had complete pre-NAC PET/CT, MRI (1.5T), radiology and pathology reports. PET and MRI data were acquired within 3 weeks of each other (average=1wk). We identified 123 aLNs of which 55 (45%) showed FDG-PET hypermetabolism. We compared: i) PET reports and pathology reports of the aLNs, ii) PET reports and MRI reports of the aLNs, iii) PET radiological scores and MR radiological scores of the aLNs.

T1-w dynamic contrast enhanced (DCE) MRI and PET/CT images of the same aLNs were identified with careful assessment of anatomical landmarks (Figure 1). aLNs on MRI and PET/CT images were independently scored in a blinded manner as 0,1,2, and 3 as no, low, moderate, and high likelihood of cancer in the aLNs. aLNs on MRI were considered abnormal if there were absence of fatty hilum, cortical thickening and cortical enlargement. aLN on PET were considered abnormal if they were hyperintense. Visual representative nodes of each scoring value are provided in Figure 2. Analyses was done under the guidance of breast radiologists.

Results

The agreement between PET reports versus needle-biopsy pathology reports on presence of metastasis in aLNs was 87% (39 out of 45 patients), indicating that PET reports can be used as the reference standard for comparison of matched aLNs. Similarly, the agreement between PET reports and MRI reports was 93% (42 out of 45 patients). Representative PET and MRI of the nodes with and without disease, as determined by pathology, are shown in Figure 3.

The results of individual scores are shown in Figure 4, with green data points being those of agreement and red data points being non-agreement. There was good agreement between MRI and PET scores (Cohen’s Kappa 𝜅= 0.39, p=10-5). The MRI sensitivity, specificity, NPV and accuracy were 66.1%, 72.1%, 73.1% and 69.9%, respectively.

Discussion

There was good agreement between MRI and PET scores. These results are consistent with a few FDG-PET and MRI studies assessing metastasis in the aLNs. FDG-PET diagnostic accuracy of pathologically confirmed disease in the aLN was 76.4% [3], FDG-PET detection sensitivity was 61% and specificity 80% [6], and FDG-PET detection sensitivity was 62.5% sensitivity and specificity 91.3% [7].

Using standard protocol covering both the breast and axilla, previous studies showed MRI sensitivity was 82.0% and negative predictive value (NPV) was 82.6% [4]. With a dedicated axillary protocol, MRI sensitivity to disease in the aLN was 84.7% and NPV was 95.0% [4,8]. DCE-MRI had a lower median sensitivity (60.0%) and NPV (80.0%) compared to non-enhanced T1-w and T2-w MRI sequences (88.4, 94.7%), and diffusion-weighted imaging (84.2, 90.6%) [4]. When comparing across studies, it is important to note whether or not the cohorts were enriched by diseased nodes or not. It is easier to score obviously diseased nodes and thus sensitivity and specificity might dependent on the data cohort. In our cohort, about half (45%) the nodes were hypermetabolic as assessed on PET.

Conclusion

Acknowledgements

References

[1] Maxwell, F., C. de Margerie Mellon, M. Bricout, E. Cauderlier, M. Chapelier, M. Albiter, P. Bourrier, M. Espié, E. de Kerviler, and C. de Bazelaire. 2015. "Diagnostic Strategy For The Assessment Of Axillary Lymph Node Status In Breast Cancer". Diagnostic And Interventional Imaging 96 (10): 1089-1101. doi:10.1016/j.diii.2015.07.007.

[2] Harbeck, N. and Thomssen, C. 2011 . A New Look at Node-Negative Breast Cancer. The Oncologist , vol. 16 Supplement 1 51-60.

[3] Ahn, Jhii-Hyun, Eun Ju Son, Jeong-Ah Kim, Ji Hyun Youk, Eun-Kyung Kim, Jin Young Kwak, Young Hoon Ryu, and Joon Jeong. 2010. "The Role Of Ultrasonography And FDG-PET In Axillary Lymph Node Staging Of Breast Cancer". Acta Radiologica 51 (8): 859-865. doi:10.3109/02841851.2010.501342.

[4] Kuijs, V. J. L., M. Moossdorff, R. J. Schipper, R. G. H. Beets-Tan, E. M. Heuts, K. B. M. I. Keymeulen, M. L. Smidt, and M. B. I. Lobbes. 2015. "The Role Of MRI In Axillary Lymph Node Imaging In Breast Cancer Patients: A Systematic Review". Insights Into Imaging 6 (2): 203-215. doi:10.1007/s13244-015-0404-2.

[5] Valente, Stephanie A., Gary M. Levine, Melvin J. Silverstein, Jessica A. Rayhanabad, Janie G. Weng-Grumley, Lingyun Ji, Dennis R. Holmes, Richard Sposto, and Stephen F. Sener. 2012. "Accuracy Of Predicting Axillary Lymph Node Positivity By Physical Examination, Mammography, Ultrasonography, And Magnetic Resonance Imaging". Annals Of Surgical Oncology 19 (6): 1825-1830. doi:10.1245/s10434-011-2200-7.

[6] Wahl, Richard L., Barry A. Siegel, R. Edward Coleman, and Constantine G. Gatsonis. 2004. "Prospective Multicenter Study Of Axillary Nodal Staging By Positron Emission Tomography In Breast Cancer: A Report Of The Staging Breast Cancer With PET Study Group". Journal Of Clinical Oncology 22 (2): 277-285. doi:10.1200/jco.2004.04.148.

[7] Groves, Ashley M, Manu Shastry, S. Ben-Haim, I. Kayani, A. Malhotra, T. Davidson, and T. Kelleher et al. 2012. "Defining The Role Of PET-CT In Staging Early Breast Cancer". The Oncologist 17 (5): 613-619. doi:10.1634/theoncologist.2011-0270.

[8] Li, Chuanming, Shan Meng, Xinhua Yang, Jian Wang, and Jiani Hu. 2014. "The Value Of T2* In Differentiating Metastatic From Benign Axillary Lymph Nodes In Patients With Breast Cancer - A Preliminary In Vivo Study". Plos ONE 9 (1): e84038. doi:10.1371/journal.pone.0084038.