To characterise MR imaging biomarker in the peri-tumoral area with recurrence, progression or pseudo-progression on preoperative multi-model MRI.We retrospectively included 26 newly diagnosed cerebral glioblastoma patients, of which five showed also pseudoprogression earlier. All patients were treated with 5-aminolevulinic acid (5-ALA) fluorescence guidance with the aim of maximal tumour resection and standard postoperative concomitant chemoradiotherapy. Preoperative MRI data acquisition was performed using a 3.0 Tesla Siemens MR Magnetron system (Siemens Healthcare, Munich, Germany). Imaging included an anatomical 3D contrast T1-weighted sequence with fat suppression, perfusion, DTI, spectroscopy. We further decomposed the diffusion tensor into isotropic (p) component and anisotropic (q) component after eigenvalues were calculated in the DTI data.1 Follow up MRI at time point of progression or pseudoprogression were co-registered to the preoperative T1 with contrast by using a previously described two stage semi-automatic non-linear coregistration methods (Figure 1-A, B, C). The site of tumour progression and pseudoprogression (Figure 1-D, E) were selected based on T1 post contrast. As tumour progression and pseudoprogression both occur mainly in the periresectional area, a periresectional area of 5mm, 10mm, 15 mm and 20mm excluding the progression area was used as non-progression regions (Figure 1-F) for comparison. In addition, a contralateral area of normal appearing with matter of 5mm in diameter was used as control representing normal peritumoural brain and normal brain tissue, respectively.Patients’ clinical characteristics were shown on table 1. Patients with pseudoprogression demonstrated a better progression free survival than patients with tumour recurrence (287 ± 190 and 773 ± 272 days, respectively; p < 0.001). A similar result was seen for overall survival (534 ± 293 and 864 ± 298 days; p = 0.047). Multimodal MRI characterisation FLAIR In areas of later progression, there was significant increase in the FLAIR signal compare to all non-progression area (p = 0.0003 – 0.0024) and control (p < 0.0001). No difference between area of progression, lesion and pseudo-progression. In area of pseudo-progression, there was increased signal in FLAIR than non-progression area and control, but this achieve only marginal significance. Diffusion MR Characterising isotropic p component (Figure 3A) showed a significant decrease in progression area than contrast enhanced area (p=0.0004), non-progression 5-15mm peritumoral area (p=0.0001 – 0.0067) and control (p<0.0001). No significant decrease in area of pseudo-progression was shown. Anisotropic q component (Figure 3D) was lower in contrast enhanced area than progression, non-progression and control (p<0.0001). In area of progression, q was lower than non-progression peritumoral 5 mm area (p=0.0204) and control (p=0.0088). Q was also lower in pseudo-progression area than the 5mm peritumoral non-progression area (p=0.0309). ADC (Figure 3B) was significant lower within the contrast enhanced area than all peritumoral area (p<0.0001). The ADC value in the progression area are lower than in the 5 mm peritumoral non-progression area (p=0.0038), but did not have difference when comparing to a more distant non-progression area. In area of pseudo-progression, the ADC value had no difference from others. Fractional anisotropy (FA) (Figure 3E) was significant lower in contrast enhanced area (p<0.0001) and was also lower in progression and pseudo-progression area when comparing to contralateral control (p=<0.0001, p=0.0129 respectively). However, FA showed no difference between progression area and non-progression area (p=0.7041-0.8390). DSC-MRI Relative cerebral blood volume was increased in lesion, progression area and pseudo-progression area comparing to control (Figure 3C, p=0.002, <0.0001 and 0.041 respectively). However, there was no difference between non-progression area and progression area or pseudo-progression area. MRS MR spectroscopy data included regions of interests those occupied more than 1/2 of MRS voxel. Only 9 patients with valid MRS data were analysed (Figure 3F). Although no statistics significance were noted between progression, non-progression and control area, a trend of higher Cho/Naa, higher Choline and lower NAA+NAAG can been observed in progression area.Diffusion tensor image has been used to evaluate tumor infiltration and showed optimal results. We have previously shown that specific DTI pattern can predict this microscopic tumor invasion.5 In particular, regions with >12% decrease in q (anisotropic) component showed white matter disruption by cancer. Therefore, in our result, compare to non-progression area, a decrease in q can lead to more malignant sequel of progression. Other study using DTI showed that by calculating the relationship of axial-diffusivity and radial diffusivity, tumor infiltrated edema can be distinguished from pure vasogenic edema.2 Therefore, a change in anisotropic diffusion can be seen in tumor infiltration.Multimodal MR imaging, especially diffusion tensor imaging can demonstrate distinct characteristics in areas of potential later progression on preoperative MRI. This provides a direction for future study and treatment target.