[Quantitative analysis of diffusion-weighted magnetic resonance images during chemoradiation therapy for cancer of the cervix uteri: Prognostic role of pretreatment diffusion coefficient values].

OBJECTIVE

to carry out a quantitative analysis of diffusion-weighted magnetic resonance images (DWI) in cancer of the cervix uteri (CCU) and to estimate the possibility of using pretreatment measured diffusion coefficient (MDC) to predict chemoradiation therapy (CRT).

MATERIAL AND

METHODS

The investigation prospectively enrolled 46 women with morphologically verified Stages IB-IVB CCU. All the women underwent diffusion-weighted magnetic resonance imaging of pelvic organs before and after treatment. A semiautomatic method was used to determine tumor signal intensity (SI) on DWI at b 1000 s/mm2 (SI b1000) and tumor MDC. The reproducibility of MDC measurements was assessed in 16 randomly selected women. The investigators compared the pretreatment quantitative DWI measures in complete and incomplete regression (CR and IR) groups and the presence and absence of tumor progression during a follow-up. An association of MDC with progression-free and overall survivals (PFS and OS) was determined in the patients.

RESULTS

A semiautomatic tumor segmentation framework could determine the pretreatment quantitative DMI measures with minimal time spent and high reproducibility. The mean tumor MDC was 0.82 +/- 0.14 x 10(-3) mm2/s. CR and IR were established in 28 and 18 women, respectively. The MDC < or = 0.83 x 10(-3) mm2/s predicted CR with a sensitivity of 64.3% and a specificity of 77.8% (p=0.007). The median follow-up was 47 months (range, 3-82 months). With the MDC < or = 0.86 x 10(-3) mm2/s, 5-year PFS was 74.1% versus 42.1% with a higher MDC (p=0.023) and 5-year OS was 70.4 and 40.6%, respectively (p=0.021). The survival difference was insignificant in relation to the degree of tumor regression. The pretreatment IS at b1000 was of no prognostic value.

CONCLUSION

The pretreatment tumor MDC may serve as a biomarker for predicting the efficiency of CRT for CCU.