Breast MRI

Last revised by Joshua Yap on 24 Apr 2023

Breast MRI is the most sensitive method (>90%) for the detection of breast cancer. Its role in diagnosis and management continues to evolve 13.

Dynamic contrast-enhanced (DCE)-MRI provides information about the morphology and function of a lesion with high sensitivity but moderate specificity (72%) which leads to subsequent patient work-up and biopsies with results that indicate benignity. As a consequence, other MRI sequences, in addition to DCE-MRI images, have been introduced in routine breast MRI exams. This approach has been defined as multiparametric MRI (mpMRI), in which T2-weighted MRI is frequently used. 14

69% of breast cancer detected on MRI are smaller than 1 cm, and the mean size is about 0.8 cm. This earlier detection leads to downstaging of the average cancer and improves breast cancer‐specific survival.

10-15% of DCIS present as non-calcifying DCIS and are undetected on mammogram but detected on MRI. 70 -80% of cases of DCIS on MRI present as non-mass enhancement in a ductal or segmental distribution with a clumped or stippled morphological appearance. The remaining 20-30% present various enhancement patterns such as focus or a mass in a focal area or regional distribution.15

  • characterization of lesions

  • discrimination between benign and malignant breast lesions

  • preoperative staging

  • tumor size estimation

  • detection of the invasive component in DCIS lesions  

  • detection of additional tumor foci in the ipsilateral and contralateral breast

  • to improve breast cancer surgery

  • to document response in patients treated with neoadjuvant chemotherapy

  • abbreviated MRI protocols are being used for screening purposes in high-risk patients like those with BRCA gene mutation showing that it detects most breast cancers at an early stage 13

  • superior to mammography when assessing patients with dense breast parenchyma to detect additional occult cancer foci

  • dynamic T1-weighted fat sat gradient-echo before and after IV gadolinium injection

  • T2W-TSE or STIR sequences

  • diffusion-weighted imaging (DWI)

  • further techniques, e.g. proton MR-spectroscopy, seldom applied outside research settings

  • amount of fibroglandular tissue

    • almost entirely fat: ACR a

    • scattered fibroglandular tissue: ACR b

    • heterogeneous fibroglandular tissue: ACR c

    • extreme amount of fibroglandular tissue: ACR d

  • background parenchymal enhancement (BPE)

    • minimal

    • mild

    • moderate

    • marked

Enhancing lesions are common. The absence of enhancement practically excludes breast cancer with a negative predictive value (NPV) >99%.

  • focus (formerly defined as <5 mm): enhancement too small to be characterized, should be considered as BPE if symmetric and multiple

  • mass enhancement (space-occupying lesion, best diagnostic clue: margins can be assessed)

    • shape: round (non-specific), oval (rather benign), irregular (rather suspicious)

    • margins: circumscribed (benign), non-circumscribed (rather suspicious), spiculated (highly suspicious)

    • internal enhancement pattern: homogeneous (rather benign), heterogeneous (non-specific), rim (rather suspicious, in particular, if centripetal, filling up over time), dark internal septations (rather benign), old BI-RADS included central enhancement (part of the lesion enhances, highly specific for fibroadenoma)

  • non-mass enhancement (best diagnostic clue: margins cannot be assessed due to diffuse enhancement or grouped multiple spots; non-mass are far more difficult to distinguish and reflect different pathological entities)

    • distribution pattern: focal, linear (rather suspicious), regional, multiple regional, segmental (rather suspicious)

    • internal enhancement pattern: homogeneous (rather benign), heterogeneous (non-specific), clumped (rather suspicious), clustered ring (rather suspicious, seldom seen), old BI-RADS included stippled, a homogeneous grainy enhancement typically benign

  • in all lesions: enhancement kinetics: see breast MRI enhancement curves

    • washout (rather suspicious), plateau (non-specific), persistent (rather benign) (caveat: lymph nodes show washout but typical morphology)

  • BIRADS 0: incomplete/non-diagnostic - this category should not be used for marked background parenchymal enhancement (BPE), motion artifacts etc.

  • BIRADS I: negative (no enhancing lesions, no benign changes such as scars, cysts etc.)

  • BIRADS II: benign (lymph nodes, inflamed cysts, fibroadenoma, fat necrosis, foci/stippled enhancement, patchy BPE).

  • BIRADS III: probably benign, requiring short-term follow-up in 6 months. If the finding is visible (e.g. on ultrasound), the most widely available method should be used for follow-up (should be applied only to lesions not fitting category II and IV, probably benign findings in high-risk screening should rather be biopsied than followed-up)

  • BIRADS IV: suspicious finding requiring biopsy (biopsy should always be tried by ultrasound first as the majority of MRI lesions can be localized by targeted ultrasound)

  • BIRADS V: highly suspicious, biopsy mandatory

  • BIRADS VI: known, histologically-verified cancer

Diagnosis is established by combining morphological and functional criteria. A circumscribed round lesion with persistent enhancement is a typical fibroadenoma while the same lesion presenting with washout may be cancer (typical in the high-risk population, aggressive cancers).

Clinical history and correlation with mammography are not only diagnostically useful (e.g. to reduce the number of BIRADS III category assignments) but should be considered in the report as well in order to demonstrate to the referring physician that the clinical question has been answered.

  • in high-risk screening population: 3-4% prevalence when mammography was negative ( 0.3% when mammography and ultrasound negative)

  • 7% if a personal history of cancer

  • positive predictive value 24% (½ invasive 4 mm median size / ½ DCIS)

    • biopsy recommended in 17%

  • contralateral breast

    • 5% prevalence

    • 20% positive predictive value (biopsy recommended in 1/3) (NEJM 29/3/2007: biopsy recommended in 12% PPV 25%);

  • ipsilateral breast: ~25%

  • 50% PPV (biopsy recommended in 50%)

  • ipsilateral multifocal ¾ (same quadrant >1 cm from index CA or contiguous but extends >4 cm) multicentric ¼; distribution similar to recurrent disease

  • additional sites of ipsilateral cancer more frequent if +FH (42%) & ILC (55%)

  • positive predictive value higher the closer the lesion is to the index cancer.

  • biopsy to get histological diagnosis no matter how suspicious because the result is mastectomy

  • younger patients because of 1-2% per year recurrence may also benefit from preoperative MRI

  • true and false positive rates decrease with each subsequent comparison MRI

  • high-risk screening

  • extent of disease (EOD) evaluation in ipsilateral and contralateral breast

  • positive margins (better accuracy further from lumpectomy site than near lumpectomy site due to postoperative changes)

  • neo-adjuvant chemotherapy: to assess residual disease

  • metastatic axillary lymphadenopathy of unknown primary (75 - 80% sensitive) - can spare a patient from having management because may be able to undergo BCT; management path only finds cancer in two-thirds

  • posterior lesion to assess chest wall invasion (pectoralis can be resected so not considered

  • chest wall stage IIIB - serratus anterior muscle, rib, intercostal muscles)

  • BRCA+ : BRCA1 or BRCA2

  • first-degree relative BRCA+ and untested

  • those who have had prior radiotherapy to chest wall 

  • >25% lifetime risk based on genetic models (some of which take breast density into consideration)

  • not recommended if lifetime risk <15% because of high false-positive rate

  • problem-solving (e.g. post-operative breasts with distortion)

  • recurrent breast cancer/scar changes (not usual before 2 - 3 years; peak 5 - 7 years; increased risk if EIC, younger age, positive margins (wait at least 1 month postop to scan), no radiotherapy)

  • to assess for synchronousmultifocal or multicentric disease

Features with the highest predictive value for carcinoma: spiculated margins (100%) and irregular shape (97%) 17.

In non-masslike enhancement lesions, the ductal enhancement possesses the highest PPV (0.500) follow by clumped enhancement (PPV, 0.304) 17.

With regards to the kinetic evaluation of breast MR lesions, type II curve or plateau pattern and type III curve is also known as a washout pattern, are considered concerning for malignancy and strongly suggestive of malignancy respectively. See the main article on breast MRI enhancement curves.

  • BI-RADS category 5 lesions have a PPV of 0.714 

  • BI-RADS category 4 lesions have a PPV of 0.205 17

  • BI-RADS category 1 and 2 lesions have an NPV of 99%

  • calcifications

    • segmental: 67%

    • clumped ductal: 31%

  • malignant causes: DCIS, invasive cancer

  • benign high-risk causes: ADH, LCIS

  • benign: fibrosis, ductal hyperplasia, fibrocystic change

  • 40-50% of cancers should be <1 cm

  • at least 20-30% should be DCIS

  • positive nodes <20%

  • technical causes: breast tissue not included in the coil, motion, bad contrast injection, too much compression

  • marked background enhancement

An important caveat is that if mammography or ultrasound is positive or there is a palpable finding then there is a need to treat/biopsy/excise despite negative MRI.

  • MSKCC: only 23% probably low but if the lesion is less than 1 cm or deep within lots of background parenchyma in a large breast may want to go directly to MR guided biopsy

  • breast MRI is better than clinical assessment, mammogram, and ultrasound in correlation with pathology; pretreatment MRI should be performed and compared with MRI done after neoadjuvant chemotherapy, this is particularly useful to monitor response to neoadjuvant chemotherapy allowing to identify non-responders early and to delineate the residual tumor after neoadjuvant chemotherapy to determine the appropriate extent of surgical excision 15

  • in patients with a personal history of breast cancer, MRI sensitivity ranges from 80–100%, whereas mammography sensitivity ranges from 0–53%; specificities for MRI screening in this population are relatively high but still lower than for mammography 16

  • MRI performed in extremely dense breasts has an additional cancer detection of 3.6% in patients with negative mammography ref

ADVERTISEMENT: Supporters see fewer/no ads

Cases and figures

  • Case 1: breast cancer and lipoma
    Drag here to reorder.
  • Case 2: invasive ductal carcinoma
    Drag here to reorder.
  • Case 3: invasive lobular carcinoma
    Drag here to reorder.
  • Case 4: extra-capsular implant rupture
    Drag here to reorder.
  • Case 5a: multicentric lobular cancer
    Drag here to reorder.
  • Case 5b: color coded image
    Drag here to reorder.
  • Case 6a: pseudoangiomatous stromal hyperplasia (PASH)
    Drag here to reorder.
  • Case 6b: enhancement curve
    Drag here to reorder.
  • Case 7: high grade ductal carcinoma in situ (DCIS)
    Drag here to reorder.
  • Updating… Please wait.

     Unable to process the form. Check for errors and try again.

     Thank you for updating your details.