Does breast magnetic resonance imaging measurement correlate with pathology in assessment of primary breast cancer?

Introduction

Magnetic resonance imaging (MRI) has been increasingly used for diagnosis and staging of breast tumours in selected clinical scenarios. The aim of our study was to compare MRI and histological measurement of primary breast lesions.

Methods

Retrospectively, patients with breast cancer who underwent MRI of the breast between 2006 and 2009 were identified from the Radiology Information System database. The maximum dimension of breast lesion on MRI was recorded by two breast radiologists and all lesions were 'T' staged. Correlation between MRI and histological measurement was performed using SPSS version 14.0. Measurement agreement by MRI and histology was analysed using the Bland and Altman (B&A) plot.

Results

This study included 98 patients. The mean size on MRI was 27 mm and that on histology was 28.87 mm. On the basis of MRI, 43 cases were classified as T1, 48 as T2 and 7 as T3. There was significant correlation between the MRI and histological measurements, with a correlation coefficient of 0.770 (P = 0.0001) and R² value of 0.594. On B&A analysis the arithmetic mean difference (AMD) between MRI and histological measurements was 1.86 mm but the limits of agreement (LOA) were -24.5 to 28 mm. T stage comparison on B&A plot showed an AMD of -4.6 mm (LOA -28 to 18.9 mm) for T1, 1.25 mm (LOA -22.9 to 25.4 mm) for T2 and -6.42 mm (LOA -53 to 40 mm) for T3.

Conclusion

A high degree of correlation exists between breast MRI and histological measurement using correlation coefficients. However, on B&A plot there is significant over and underestimation of T stage by MRI.