Introduction
Materials and methods
Search strategy
In- and exclusion criteria
Study selection
Data extraction and quality assessment
Results
Publication characteristics
Technical MRI details
Reference test
Technical MRI details | |||||||||
---|---|---|---|---|---|---|---|---|---|
Author | Year | Study design | Field strength | Dedicated vs. covering* | Coil | Imaging sequences | Contrast used | Voxel size | Image analysis |
Yoshimura et al. | 1999 | P | 1.5 T | Dedicated | 3-Inch circular SC | T1w | None | 1.3 × 0.6 × 3.0 mm3 | Short and long axis ratio and morphology |
Kvistad et al. | 2000 | P | 1.5 T | Dedicated | BC and extension to axilla | T1w, dynamic T1w | Gadodiamide | 2.0 × 1.0 × 4.0 mm3 | Normal diameter 2–3 mm, signal intensity time curves, morphology |
Michel et al. | 2002 | P | 1.5 T | Dedicated | Cardiac SC | T1w, T2w, T2*w | Ferumoxtran-10 | T1w 0.9 × 0.9 × 3.5, T2w 0.4 × 0.4 × 3.5, T2*w 0.9 × 0.9 × 2.0 mm3 | Size, shape, long-shirt axis ratio, USPIO uptake patterns |
Harada et al. | 2007 | P | 1.5 T | Dedicated | Cardiac surface coil | T1w, T2*w, dynamic T2*w | Ferumoxtran-10 | 0.63 × 1.25 × 0.5 mm3 | Pre contrast: short axis >5 or >10 mm, replacement fatty hilum, irregular margins. USPIO: heterogeneous uptake or lack of uptake |
Orguc et al. | 2012 | P | 1.5 T | Breast and axilla | Not reported | T2w, dynamic T1w | Not reported | Not reported | Size, SI time curves |
Fornasa et al. | 2012 | P | 1.5 T | Breast and axilla | 4-Channel phased array coil | T1w, T2w, dynamic T1w, DWI (b-value 0/800) | Gadoterate meglumine | T2w 1.3 × 1.3 × 4.0; T1w 1.1 × 1.3 × 4.0; dynamic T1w 1.4 × 1.4 × 1.8; DWI 5.3 × 5.3 × 4.0 mm3 | Lymph node area, ADC |
He et al. | 2012 | P | 1.5 T | Breast and axilla | 8-Channel BC | T1w, dynamic T1w, T2w, DWI (b-value 0/500/800) | Gadopente-tate dimeglumine | Dynamic T1w 1.1 × 1.1 × 2 mm3; DWI 2.7 × 2.7 × 4 mm3; T1w not reported (only ST: 5 mm) | SI time curve, shape, margin, diameter, long/short axis ratio, node anatomical location, ADC values, high SI on DWI, early stage enhancement |
Scaranelo et al. | 2012 | P | 1.5 T | Breast and axilla | 8-Channel BC | T1w, T2w, DWI (b-value 0/50/300/700/1000) | None | T1 0.7 × 0.9 × 1.2 mm3; (T2 not reported; DWI only ST: 4 mm) | Shape, presence of fatty hilum, cortex irregular, lobulated margins, visual inspection of DWI and ADC |
Hwang et al. | 2013 | R | 1.5 T | Breast and axilla | 4-Channel BC | T1w, T2w, DWI (b-value 750/1000), Dynamic T1w | Gadopente-tate dimeglumine | Not reported (only ST: T1w 3.4, T2w 2.6, DWI 3.4, Dynamic T1w 2.6 mm) | Cortical thickening, irregular or round shape, loss of fatty hilum |
Luo et al. | 2013 | P | 1.5 T | Breast and axilla | 8-Channel BC | T1w, T2w, DWI | None | T1w 1.0 × 0.7 × 1.0; T2w 1.1 × 1.1 × 4.03; DWI 1.0 × 0.7 × 6.0 mm3 | Lymph nodes >4 mm, ADC values |
Kamitani et al. | 2013 | R | 1.5 T | Breast and axilla | Body coil | T1w, dynamic T1w, T2w, DWI | |||
(b-Value 0/1000) | Gadopente-tate dimeglumine | T1w 0.93 × 0.93 × 4.0 mm3; DWI 2.8 × 4.1 × 5.0 mm3 (T2w not reported) | Short axis, detectability on DWI | ||||||
Basara et al. | 2013 | P | 1.5 T | Breast and axilla | 8-Channel BC | CE-T1w, T2w, DWI | |||
(b-Value 0/600) | Not reported | DWI: 0.2 × 0.2 × 5.0 mm3; other sequences not reported | Size, ADC | ||||||
Hieken et al. | 2013 | R | 1.5 T | Breast and axilla | BC | Dynamic T1w | Gadobenate dimeglumine | Not reported | Cortical thickness >3 mm, asymmetric cortex, shape, unclear margins, perinodular oedema, loss or displacement of hilum, matting |
Abe et al. | 2013 | P | 1.5 T | Breast and axilla | BC | T2w, dynamic T1w | Gadodiamide | 1.0 × 1.0 mm2, ST not reported | Diffuse and asymmetrical cortical thickening, loss of hilum |
Li et al. | 2014 | P | 3.0 T | Dedicated | 12-Channel body coil | T1w, T2w, T2*w | None | 0.8 × 0.8 × 3.0 mm3 | T2* values |
An et al. | 2014 | R | 1.5 or 3.0 T | Breast and axilla | BC | T2w, T1w, dynamic T1w | Gadobutrol | Both 1.5T and 3.0T; T2w ST 3 mm, dynamic unenhanced and CE T1w FOV 320 × 320, ST 3 mm | Loss of fatty hilum, cortex thickness of >3 mm, irregular or round shape on T2w |
Author | LN imaging analysis | Time to surgery (days) |
n
| Prevalence N+ | Mean age in years (range) | T-stage at diag-nosis | Histologic type | Pathologic analysis |
---|---|---|---|---|---|---|---|---|
Yoshimura et al. | N-N | “Preoperative MRI”, n.o.s. | 202 | 40.0 % | 55 (30–86) | T1-T3 | PT, ST, scirrhous, other | Sectioned long-axis |
Kvistad et al. | P-P | “Before surgery”,n.o.s | 65 | 37.0 % | 59.4 (38–79) | T1-T4 | IDC, ILC, MC, tubular, adenocarcinoma | Histopathological examination n.o.s. |
Michel et al. | P-P and N-N | 2 days (range 1–6 days) | 18 | 61.1 % | 53 (22–76) | T1-T4 | NR | LN regarded as positive when tumour cells were present at light microscopy, independent from immunohistochemical staining results. |
Harada et al. | N-N | Mean 1.2 days | 33 | 19.0 % | 58 (36–77) | T1-T4 | PT, ST, scirrhous, medullary, MC, apocrien, SC, spindle cell | Sectioned long-axis, H&E staining, conventional microscopic examination |
Orguc et al. | N-N | NR | 155 | 25.8 % | 43 (28–76) | NR | NR | NR |
Fornasa et al. | One LN/ALND/patient | 14 days | 43 | 44.2 % | 58 (39–78) | NR | IDC, ILC | Position of the LN and long × short axis |
He et al. | N-N | “After MRI”, n.o.s. | 79 | 12.0 % | 44 (20–67) | NR | DCIS, IDC, ILC, lymphoma | Each LN MRI removed, residual fatty tissue examined, LN sliced 4-mm sections, 3-μm-thick slices cut from each section, stained with H&E |
Scaranelo et al. | P-P | “After MRI”, n.o.s. | 61 | 43.0 % | 53 (33–78) | NR | NR | Embedded in paraffin blocks for histopathologic evaluation n.o.s. |
Hwang et al. | P-P | “After MRI”, n.o.s. | 349 | 26.4 % | 51.3 (25–79) | Only T1 | IDC, ILC, MC, others | Intraoperative frozen section, H&E staining. Remaining portions SLNB; sectioning and immunohistological assay |
Luo et al. | N-N | “Preoperative MRI”, n.o.s. | 36 | 57.0 % | 53 (30–63) | T1-T3 | DCIS, IDC, ILC, other | Analysed and examined by pathologist n.o.s. |
Kamitani et al. | P-P and N-N | NR | 108 | 23.6 % | 54.9 (34–84) | Tis-T3 | DCIS, IDC,ILC, MC, metaplastic | NR |
Basara et al. | P-P and N-N | NR | 110 | 24.0 % | Benign 47 (19–73); malignant 43 (29–70) | NR | IDC, ILC, IDC+ILC, IDC+MC, IDC+pleomorphic, medullary, malignant phyllodes | NR |
Hieken et al. | P-P and N-N | ‘Preoperative MRI”, n.o.s. | 505 | 30.1 % with N0i+, 27.3 % without N0i+ | 62 (24–91) | T1-T4 | IDC, ILC, mixed mammary carcinoma, other | Pathology was reviewed and the presence and extent of axillary nodal disease was verified n.o.s. |
Abe et al. | P-P | NR | 50 | 32.0 % | 59.9 (33–83) | T1-T3 | IDC, ILC | Pathologically confirmed with SLNB or ALND n.o.s. |
Li et al. | N-N | NR | 35 | 42.0 % | NR (30–58) | T1-T2 | IDC, ILC, tubular | Parallel slices 2–3 mm thickness and stained with H&E |
An et al. | P-P | ‘Preoperative MRI”, n.o.s. | 215 | 61.4 % | 50 (26–83) | T1-T3 | IDC, ILC, MC, invasive micro papillary carcinoma, metaplastic, medullary | Sections stained with H&E |
Quality of included studies
Diagnostic performance of MRI in axillary lymph node staging: Dedicated to the axilla versus covering the breast and axilla
First author, year | Sensitivity (95 % CI) | Specificity (95 % CI) | NPV (95 % CI) | PPV (95 % CI) | Accuracy (95 % CI) | |
---|---|---|---|---|---|---|
Dedicated to axilla | ||||||
T1w/T2w/DCE | ||||||
Yoshimura et al. 1999 | 79.0 % | 93.0 % | 87.0 % | 89.0 % | 88.0 % | |
Kvistad et al. 2000 | 83.0 % | 90.0 % | 90.0 % | 83.0 % | 88.0 % | |
Li et al. 2014 | 94.6 % | 98.5 % | 95.0 % | 98.2 % | Not reported | |
Median
|
88.4 % |
93.0 % |
94.7 % |
89.0 % |
86.5 % | |
Range
|
79–94.6 % |
90–98.5 % |
87–95 % |
83.0–98.2 % |
88.0 % | |
T2*w USPIO | ||||||
Michel et al. 2002 | Disease based | 82.0 % | 100 % | 78.0 % | 100,0 % | 89.0 % |
Two readers lymph node based | 73.0–83.0 % | 96.0–97.0 % | 97.0–98.0 % | 71.0–74.0 % | 94.0–95.0 % | |
Harada et al. 2006 | Combined study | 86.4 % | 97.5 % | 96.1 % | 91.1 % | 95.0 % |
Postcontrast alone | 84.7 % | 96.8 % | 95.6 % | 88.5 % | 94.0 % | |
Median
|
83.0 % |
97.0 % |
95.9 % |
89.8 % |
94.3 % | |
Range
|
73.0–86.4 % |
96.0–100 % |
78.0–98.0 % |
71.0–100 % |
89.0–95.0 % | |
Median
| 84.7 % | 96.8 % | 95.0 % | 89.0 % | 91.5 % | |
Range
|
73–94.6 % |
90.0–100.0 % |
78.0–98.0 % |
71.0–100 % |
88.0–95.0 % | |
Covering breast and axilla | ||||||
T1w/T2w/DCE | ||||||
Orguc et al. 2012 | 89.0 % | 14.0 % | 80.0 %* | 21.4 %* | Not reported | |
He et al. 2012 | Overall | 33.3–86.5 % | 95.2–98.2 % | 1.9–16.7 % | 66.7–82.6 % | Not reported |
Early stage enhancement rate** | 97.0 % | 73.5 % | 99.5 %* | 30.5 %* | Not reported | |
Scaranelo et al. 2012 | 88.4 % (76–95) | 82.4 % (71–90) | 94.7 %* | 69.4 %* | 85.0 % (77–91) | |
Hwang et al. 2013 | 47.8 % | 88.7 % | 82.6 % | 60.2 % | 77.9 % | |
Hieken et al. 2013 | N0 with N0i+ | 54.2 % (46.6–61.6) | 78.2 % (73.2–82.5) | 75.7 % (70.7–80.1) | 57.7 % (49.9–65.2) | 69.7 % |
N0 without N0i+ | 57.2 % (49.1–64.9) | 78.2 % (73.2–82.5) | 78.9 % (74.0–83.2) | 56.2 % (48.2–63.9) | 71.3 % | |
Abe et al. 2013 | 60.0 % | 79.0 % | 81.0 % | 59.0 % | 74.0 % | |
An et al. 2014 | 67.5 % | 78.0 % | 79.2 % | 65.9 % | 74.0 % | |
Median
|
60.0 % |
78.6 % |
80.0 % |
59.0 % |
74.0 % | |
Range
|
33.3–97.0 % |
14.0–98.5 % |
1.9–99.5 % |
21.4–92.6 % |
69.7–85.0 % | |
DWI | ||||||
Fornasa et al. 2012 | Cutoff <1.09 × 10−3 mm2/s | 94.7 % | 91.7 % | 95.7 % | 90.0 % | 93.0 % |
He et al. 2012 | Cutoff <1.35 × 10−3 mm2/s | 97.0 % | 54.5 % | 99.4 %* | 20.4 %* | Not reported |
Scaranelo et al. 2012 | No cutoff value | 83.9 % (73–91) | 77.0 % (65–86) | 90.9 %* | 60.5 %* | 80.0 % (72–86) |
Luo et al. 2013 | Cutoff <0.89 × 10−3 mm2/s | 82.2 % | 82.4 % | 77.8 % | 86.1 % | 82.3 % |
ADC ratio*** ≤1.097 | 84.4 % | 88.2 % | 81.1 % | 90.5 % | 86.1 % | |
Kamitani et al. 2013 | Cutoff <1.05 × 10−3 mm2/s | 53.8 % | 86.9 % | 85.9 % | 56.0 % | 79.1 % |
Cutoff <1.22 × 10−3 mm2/s | 75.6 % | 71.1 % | 90.2 % | 54.3 % | Not reported | |
Basara et al. 2013 | Cutoff <1.49 × 10−3 mm2/s | 95.6 % | 30.3 % | 95.6 % | 30.3 % | Not reported |
Median
|
84.2 % |
79.7 % |
90.6 % |
58.3 % |
82.3 % | |
Range
|
53.8–97.0 % |
30.3–91.7 % |
77.8–99.4 % |
20.4–90.5 % |
79.1–93.0 % | |
Median
| 82.0 % | 78.2 % | 82.6 % | 59.0 % | 79.1 % | |
Range
|
33.3–97.0 % |
14.0–98.5 % |
1.9–95.7 % |
30.3–92.6 % |
69.7–93.0 % |
Diagnostic performance of MRI in axillary lymph node staging: Studies using T1w/T2w, DCE, DWI and T2*w USPIO imaging sequences
First author, year | Prevalence N+ % | Sensitivity (95 % CI) | Specificity (95 % CI) | NPV (95 % CI) | PPV (95 % CI) | Accuracy (95 % CI) | |
---|---|---|---|---|---|---|---|
Imaging sequences T1w/T2w | |||||||
Yoshimura et al. 1999 |
40.0 % | T1w | 79.0 % | 93.0 % | 87.0 % | 89.0 % | 88.0 % |
Scaranelo et al. 2012 |
43.0 % | T1w | 88.4 % (76–95) | 82.4 % (71–90) | 94.7 %* | 69.4 %* | 85.0 % (77–91) |
Li et al. 2014 |
42.0 % | T2*w | 94.6 % | 98.5 % | 95.0 % | 98.2 % | Not reported |
Median
|
88.4 % |
93.0 % |
94.7 % |
89.0 % |
86.5 % | ||
Range
|
79.0–94.6 % |
82.4–98.5 % |
87.0–95.0 % |
69.4–98.2 % |
85.0–88.0 % | ||
Imaging sequences DCE | |||||||
Kvistad et al. 2000 |
37.0 % | 83.0 % | 90.0 % | 90.0 % | 83.0 % | 88.0 % | |
Orguc et al. 2012 |
25.8 % | 89.0 % | 14.0 % | 80.0 %* | 21.4 %* | Not reported | |
He et al. 2012 |
12.0 % | Overall | 33.3–86.5 % | 95.2–98.2 % | 1.9–16.7 % | 66.7–82.6 % | Not reported |
Early stage enhancement rate** | 97.0 % | 73.5 % | 99.5 %* | 30.5 %* | Not reported | ||
Hwang et al. 2013 |
26.4 % | 47.8 % | 88.7 % | 82.6 % | 60.2 % | 77.9 % | |
Hieken et al. 2013 |
30.1 % | N0 with N0i+ | 54.2 % (46.6–61.6) | 78.2 % (73.2–82.5) | 75.7 % (70.7–80.1) | 57.7 % (49.9–65.2) | 69.7 % |
27.3 % | N0 without N0i+ | 57.2 % (49.1–64.9) | 78.2 % (73.2–82.5) | 78.9 % (74.0–83.2) | 56.2 % (48.2–63.9) | 71.3 % | |
Abe et al. 2013 |
32.0 % | T1w DCE | 60.0 % | 79.0 % | 81.0 % | 59.0 % | 74.0 % |
An et al. 2014 |
61.4 % | 67.5 % | 78.0 % | 79.2 % | 65.9 % | 74.0 % | |
Median
|
60.0 % |
78.2 % |
80.0 % |
59.0 % |
74.0 % | ||
Range
|
33.3–97.0 % |
14.0–98.2 % |
1.9–99.5 % |
21.4–83.0 % |
69.7–88.0 % | ||
Imaging sequences DWI | |||||||
Fornasa et al. 2012 |
44.2 % | Cutoff <1.09 × 10−3 mm2/s | 94.7 % | 91.7 % | 95.7 % | 90.0 % | 93.0 % |
He et al. 2012 |
12.0 % | Cutoff <1.35 × 10−3 mm2/s | 97.0 % | 54.5 % | 99.4 %* | 20.4 %* | Not reported |
Scaranelo et al. 2012 |
43.0 % | No cutoff value | 83.9 % (73–91) | 77.0 % (65–86) | 90.9 %* | 60.5 %* | 80.0 % (72–86) |
Luo et al. 2013 |
57.0 % | Cutoff <0.89 × 10−3 mm2/s | 82.2 % | 82.4 % | 77.8 % | 86.1 % | 82.3 % |
ADC ratio*** ≤1.097 | 84.4 % | 88.2 % | 81.1 % | 90.5 % | 86.1 % | ||
Kamitani et al. 2013 |
23.6 % | Cutoff <1.05 × 10−3 mm2/s | 53.8 % | 86.9 % | 85.9 % | 56.0 % | 79.1 % |
Cutoff <1.22 × 10−3 mm2/s | 75.6 % | 71.1 % | 90.2 % | 54.3 % | Not reported | ||
Basara et al. 2013 | 24.0 % | Cutoff <1.49 × 10−3 mm2/s | 95.6 % | 30.3 % | 95.6 % | 30.3 % | Not reported |
Median
|
84.2 % |
79.7 % |
90.6 % |
58.3 % |
82.3 % | ||
Range
|
53.8–97.0 % |
30.3–91.7 % |
77.8–99.4 % |
20.4–90.5 % |
79.1–93.0 % | ||
Imaging sequences T2*w USPIO | |||||||
Michel et al. 2002 |
61.1 % | Disease based | 82.0 % | 100 % | 78.0 % | 100.0 % | 89.0 % |
Two readers lymph node based | 73.0–83.0 % | 96.0–97.0 % | 97.0–98.0 % | 71.0–74.0 % | 94.0–95.0 % | ||
Harada et al. 2006 |
19.0 % | Combined study | 86.4 % | 97.5 % | 96.1 % | 91.1 % | 95.0 % |
Postcontrast | 84.7 % | 96.8 % | 95.6 % | 88.5 % | 94.0 % | ||
Median
|
83.0 % |
97.0%
|
95.9 % |
89.8 % |
94.3 % | ||
Range
|
73.0–86.4 % |
96.0–100 % |
78.0–98.0 % |
71.0–100 % |
89.0–95.0 % |