Diffusion-weighted MRI of rectal cancer: baseline tumour perfusion fraction predicts chemoradiotherapy response and survival
Kine Mari Bakke1,2, Knut Håkon Hole3, Svein Dueland4, Krystyna Grøholt5, Kjersti Flatmark6,7,8, Anne Hansen Ree1,8, Therese Seierstad3, and Kathrine Røe Redalen1

1Department of Oncology, Akershus University Hospital, Lørenskog, Norway, 2Institute of physics, University of Oslo, Oslo, Norway, 3Department of Radiology and Nuclear Medicine, Oslo University Hospital, 4Department of Oncology, Oslo University Hospital, 5Department of Pathology, Oslo University Hospital, 6Department of Tumor Biology, Oslo University Hospital, 7Department of Gastroenterological Surgery, Oslo University Hospital, 8Institute of Clinical Medicine, University of Oslo


More accurate diagnostics for prediction of treatment responses in locally advanced rectal cancer is warranted. We employed a simplified approach to the intravoxel incoherent motion imaging method to estimate the tumour perfusion fraction from diffusion-weighted MRI. The perfusion fraction was predictive of the histologic tumour response after chemoradiotherapy (p = 0.02), and in combination with tumour volume this parameter was also predictive of five-year progression-free survival of the patients (p = 0.002). This simplified approach does not require substantial extra scan time in a routine diagnostic scanning, and may offer a clinically feasible approach to stratifying patients to individualised treatment.


Locally advanced rectal cancer (LARC) is commonly treated with chemoradiotherapy (CRT) followed by surgery1. Due to heterogeneous treatment responses to CRT, more accurate diagnostics is needed for improving patient stratification to treatment intensification, but also for identifying excellent responders for watch-and-wait approaches. Tumour hypoxia (oxygen deficiency) is acknowledged as associated with aggressive tumour development and reduced treatment response2, and this oxygen deficiency is linked to poor perfusion of the tumour. We employed a simplified approach to the intravoxel incoherent motion (IVIM) method3 to extract both the apparent diffusion coefficient (ADC) and the tumour perfusion fraction (F) from diffusion-weighted (DW) MRI. The aim was to investigate whether ADC and F predict histologic tumour regression grade (TRG) and 5-year progression-free survival (PFS) in LARC.


From baseline DWMRI of 27 patients we calculated ADC-maps using b-values 300 and 900 s/mm2. The ADC-values were used to extrapolate a line asymptotically back to the y-axis, calculating the discrepancy with the measured signal at b = 0 as F4, Figure 1. We have previously reported the predictive value of tumour volume for this study5 and we therefore included volume in our analysis to look for a potentially combined predictive value. Defining TRG1-2 as good responders and TRG3-5 as poor responders we used the Mann Whitney U test and receiver operating characteristic (ROC) curve analysis to investigate each parameters’ predictive value. Five-year survival was analysed using log rank test and Kaplan Meier plots.


ADC, though higher in good responders (74.1 x 10-4 mm2/s ± 22.7 vs 61.0 ± 12.6, p = 0.13), was not significantly predictive of TRG, nor of PFS. The perfusion fraction F, also higher in good responders (22.0 (au) ± 10.9 vs 12.6 ± 4.1, p = 0.02) was significantly predictive of TRG. The combination of F and volume (F/V) was highly predictive of TRG (AUC = 0.92, p < 0.001, CI = 0.81–1.00, sensitivity 91%, specificity 87%) and performed better than each parameter separately, Figure 2. F/V also predicted PFS, with an F/V higher than 0.5 showing a five-year PFS of 82%, compared to 30% for F/V below 0.5, Figure 3.


Using a simplified IVIM approach we identified that baseline tumor F differentiated good and poor responders to CRT and in combination with tumour volume also predicted PFS. Low F may be associated with tumour hypoxia and poor perfusion, and hence radiation resistance. A low perfusion fraction from IVIM-imaging has previously been associated with poor treatment response in locally advanced breast cancer6. ADC by itself as a predictive parameter for tumour response to CRT in LARC has not shown consistent results7, and our study shows that the additional estimate of the perfusion fraction can have clinical value. By making a simplified approach to the IVIM method, requiring only three b-values, the method does not require much extra time or heavy computation, and may therefore represent a feasible approach for implementation into clinical routine, and hence, enable improved stratification of LARC patients to individualised treatment.


No acknowledgement found.


1. GF Weber, R Rosenberg, JE Murphy et al. Multimodal treatment strategies for locally advanced rectal cancer. Expert Rev Anticancer Ther. 2012;12(4);481-94

2. RG Bristow, RP Hill. Hypoxia and metabolism. Hypoxia, DNA repair and genetic instability. Nat Rev Cancer. 2008;8(3):180-92

3. D Le Bihan, E Breton, D Lallemand et al. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology. 1988;168(2);497-505

4. R Wirestam, M Borg, S Brockstedt et al. Perfusion-related parameters in intravoxel incoherent motion MR imaging compared with CBV and CBF measured by dynamic susceptibility-contrast MR technique. Acta Radiol. 2001;42(2);123-8

5. T Seierstad, KH Hole, KK Grøholt et al. MRI volumetry for prediction of tumour response to neoadjuvant chemotherapy followed by chemoradiotherapy in locally advanced rectal cancer. BR J Radiol. 2015;88(1051)20150097

6. S Che, X Zhao, Y OU et al. Role of the Intravoxel Incoherent Motion Diffusion Weighted Imaging in the Pre-treatment Prediction and Early Response Monitoring to Neoadjuvant Chemotherapy in Locally Advanced Breast Cancer. Medicine. 2016;95(4);e2420

7. RGH Beets-Tan, GL Beets. MRI for assessing and predicting response to neoadjuvant treatment in rectal cancer. Nat. Rev. Gastroenterol. Hepatol. 2014;11;480-488


Figure 1: F was calculated as the discrepancy between the measured signal at b0 and the calculated value at b0 from an asymptotic fit from the ADC from images b300 and b900, making a simplified approach to the IVIM-method.

Figure 2: Sensitivity and specificity for the perfusion fraction F and tumour volume (V). Area under curve (AUC) is indicated.

Figure 3: The combination of the perfusion fraction (F) and tumor volume (V) divided the patients in to groups with 82% and 30% 5-year progression-free survival, respectively.

Proc. Intl. Soc. Mag. Reson. Med. 25 (2017)