Table 7.
DWI parameters during NST
| First author | Cycle (s) | ADC-value (× 10−3 mm2/s) pCR versus non-pCRa |
Reported/chosen ADC threshold for pCR (× 10−3mm2/s) | ROC AUC | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) |
|---|---|---|---|---|---|---|---|---|
| Tozaki [40] | 1 | 0.55 versus 0.78 (range 0.45–0.95) | ||||||
| Li [44] | 1 | 1.59 versus 1.24 p = 0.0019 | 1.4 | 0.82 | 83 | 67 | 59 | |
| Pereira [18] | 1 | 1.214 ± 0.0599 versus 0.954 ± 0.0267 | ||||||
| Che [19] (IVIM) | 2 | 1.36 ± 0.30 versus 0.98 ± 0.23 p = 0.001 | 0.971 | 0.851 | 100 (66.4–100) | 63.2 (38.4–83.7) | 56.3 (30.6–79.3) | 100 (69.9–99.2) |
| Zhang [24] | 2 | 1.6 ± 0.4 versus 1.1 ± 0.3 p < 0.001 | 0.864 | |||||
| Kim [53]b | 2 | ADC: 1.23 (1.10–1.38) versus 1.36 (1.32–1.57) | 1.29 | 0.70 | 0.79 | 0.62 | ||
| D: 1.15 (1.10–1.34) versus 1.37 (1.25–1.60) | 1.35 | 0.71 | 0.71 | 0.77 | ||||
| Minarikova [59]c | 2c | 1.33 ± 0.28 versus 1.13 ± 0.26 | 0.697d | |||||
| 3 & 4c | 1.24 ± 0.15 versus 1.28 ± 0.30 | 0.500 | ||||||
| Bedair [20] | 3 | 1.52 ± 0.32 versus 1.27 ± 0.18 → ADC | ||||||
| Other model-based metrics: | ||||||||
| 1.51 ± 0.15 versus 1.40 ± 0.12 → DDC | ||||||||
| 1.30 ± 0.14 versus 1.28 ± 0.15 → D | ||||||||
| 0.91 ± 0.07 versus 0.86 ± 0.11 → α (a.u.) | ||||||||
| 8.48 ± 1.54 versus 10.53 ± 2.51 → f (%) | ||||||||
| Fangberget [65]e | 4 | 1.7 (range: 1.0–2.1) versus 1.2 (range: 0.9–1.7) or 1.3 p = 0.022f | 1.42 | 88 | 80 | |||
| Minarikova [59]c | 5 | 1.10 ± 0.24 versus 1.34 ± 0.33 | 0.743 |
ADC apparent diffusion coefficient, D true diffusivity, DDC distributed diffusion coefficient, f perfusion fraction, NPV negative predictive value, PPV positive predictive value, pCR pathologic complete response, ROC AUC area under the receiver operating characteristic curve, TN triple negative, α intravoxel heterogeneity index
aMean ADC-value ± SD with the exception of Li et al. [44]: median ADC
bMiller and Payne grade 4 included as good responders
cAfter 2 cycles: 14 lesions; after 3–4 cycles: 19 lesions and 5 cycles: 34 lesions scanned
dAbout the data (ADC) and ROC-analysis: “smaller values were considered positive for pCR prediction in BS, after three to four cycles and after five to eight cycle; however, higher values were considered positive for pCR prediction in data measured after two cycles” [59]
e31 MRI at pre NAC and after 4 cycles 27 MRI’s
f1.2 × 10−3 mm2/s and 1.3 × 10−3 mm2/s for non-pCR mentioned in [65]