Purpose: Recent publications have reported contradictory results of pretreatment diffusion-weighted magnetic resonance imaging (DWI) for the prediction of chemoradiotherapeutic response in primary squamous cell carcinomas of the head and neck (HNSCC). Therefore, we evaluated the diagnostic performance of DWI obtained with both standard (b = 0 and 1,000 s/mm²) and high (b = 0 and 2,000 s/mm²) b-values for predicting response to induction chemotherapy in HNSCCs.
Methods: For 25 patients with primary HNSCC who underwent DWI with both standard and high b-values prior to treatment, we calculated corresponding apparent diffusion coefficient (ADC) maps. Regions of interest containing the tumor were drawn on every section of ADC maps and summated to make volume-based data of the entire tumor. Histogram parameters (mean ADC, kurtosis, and skewness) were correlated with treatment response using unpaired Student t test. Univariate and multivariate analysis of the ADC parameters, patient age, sex, whole tumor volume, and T stage were also performed to predict tumor response to induction chemotherapy.
Results: Response to induction chemotherapy was good in 13 of the 25 patients and poor in 12. The mean ADC values of good responders at standard b-value (ADC1000), 1.23 ± 0.34 (× 10⁻³ mm²/s), and high b-value (ADC2000), 0.62 ± 0.14 (× 10⁻³ mm²/s), were lower than those of poor responders (ADC1000, 1.32 ± 0.28 [× 10⁻³ mm²/s]; ADC2000, 0.76 ± 0.15 [× 10⁻³ mm²/s]), but significant difference was achieved only at the ADC2000 map (P = 0.02). In addition, mean tumor volume prior to treatment of good responders was smaller than that of poor responders. However, at multiple logistic regression analysis, only the mean ADC2000 value remained as a significant predictor of response to induction chemotherapy.
Conclusion: DWI with high b-values (b = 0 and 2,000 s/mm²) as an assessment of ADC values may help predict tumor response to neoadjuvant chemotherapy for primary HNSCCs.