Prediction of response by DCE-MRI and DW-MRI for intrahepatic cholangiocarcinomas treated with locoregional and systemic chemotherapy: a preliminary analysis
Kristen L Zakian1, Richard K Do2, Taryn Boucher2, Mithat Gonen3, Andrea Cercek4, William R Jarnagin5, and Nancy Kemeny4

1Medical Physics, Memorial Sloan-Kettering Cancer Center, New York, NY, United States, 2Radiology, Memorial Sloan-Kettering Cancer Center, New York, NY, United States, 3Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, New York, NY, United States, 4Medicine, Memorial Sloan-Kettering Cancer Center, New York, NY, United States, 5Surgery, Memorial Sloan-Kettering Cancer Center, New York, NY, United States

Synopsis

Intrahepatic cholangiocarcinoma (ICC) is the second most common primary liver malignancy and has few treatment options. A previous study suggested that MRI may help identify patients likely to benefit from hepatic arterial infusion pump therapy with floxuridine (HAI-FUDR). The purpose of this prospective study was to investigate the ability of pre-treatment and early-in-treatment DCE and DW-MRI to predict ICC response to combined HAI-FUDR and systemic chemotherapy given in a Phase 2 clinical trial. Our preliminary analysis suggests that DW-MRI may predict response of unresectable ICC using data acquired at baseline or at 1 month after treatment start.

Introduction

Intrahepatic cholangiocarcinoma (ICC) is the second most common primary liver malignancy and has few effective treatment options. A previous study suggested that pretreatment MRI may be helpful in identifying patients likely to benefit from hepatic arterial infusion pump therapy with floxuridine (HAI-FUDR) (1). The purpose of this prospective study was to investigate the ability of pre-treatment and early post treatment DCE- and DWI-MRI to predict response in a phase 2 trial of combination HAI-FUDR and systemic chemotherapy with gemcitabine/oxaliplatin (GemOx) for patients with unresectable ICC.

Methods

18 consecutive patients were recruited for this institutional review board approved prospective study between January 2013 and August 2015. Scans were performed at 1.5T (G.E. Signa 450, Waukeshau, WI). DCE-MRI and DWI were performed before and after treatment (1 month of HAI-FUDR and 2 weeks of GemOx). 2 patients were excluded, one with an incomplete MRI data set, one with a single tumor that was too small for analysis. 3 patients did not meet the 3 month follow-up period for response assessment. For DCE-MRI a free-breathing, 3D-fat-saturated FLASH sequence was prescribed in the oblique coronal plane (matrix 256x128, FOV 340-440, slice thickness = 5-7 mm TE = 1.5 ms, TR = 4.2 ms, flip angle 30°, temporal resolution = 6-8s). Images were prescribed over the largest tumor and Gd-BOPTA at 0.1 mmol/kg was injected at 2 ml/sec. For DWI an oblique coronal single-shot EPI sequence was used (matrix 128x128, FOV 340-440 mm, slice thickness 5-7 mm, TE = 68 ms, TR ~4s b values = 0, 50, 250, 350, 500 s/mm2). Ktrans was calculated following QIBA guidelines2. An average pre-contrast tumor T1 value was calculated based on data from 5 patients. An average AIF was modeled based on Parker, et. al.3 using data from an aortic ROI in 8 patients. In the extended Tofts’s model a single-input vascular function was utilized under the assumption that ~80% of tumor supply was arterial. ADC was calculated by monoexponential fit using all b values. Response (% change in sum of diameter) was measured after 3 months and 6 months of treatment by RECIST 1.1 on follow-up MRI. We estimated the rank correlation between Ktrans and ADC at baseline, as well as change from baseline, and response at 1, 3 and 6 months. In a supplementary analysis we dichotomized response and estimated the area under the ROC curve for each of the imaging parameters at baseline and change from baseline.

Results

At the time of analysis, 13 patients (3 males, 10 females, median age 66 years old) had measurable response at 3 months post treatment and 9 patients had measurable response at 6 months. Mean tumor ADC and median Ktrans are discussed herein. Baseline ADC was positively correlated with % change in sum of diameters (%ΔSD) at 3 and 6 months (p < 0.005 and p < 0.05, respectively), while baseline Ktrans appeared to be inversely related to %ΔSD but this was not significant (Figs 1a-d). ΔADC at 1 month was negatively correlated with %ΔSD at 3 months (p < 0.001) but was not significant for 6 month %ΔSD, while the change in Ktrans at 1 month follow-up was not correlated with %ΔSD at 3 or 6 months (Figs 2a-d). 3 patients had a partial response (PR) at 3 months (>30% %ΔSD reduction) and 6 patients had PR at 6 months. Figure 3 contains the areas under the receiver operator characteristic curves for baseline and Δparameter values for predicting PR at 3 and 6 months. Baseline ADC and ΔADC (BL-1MO) were the strongest predictors of PR.

Conclusions

Our preliminary analysis shows the potential of multimodality MRI, in particular, DW-MRI, to predict response of unresectable ICC to combination HAI-FUDR and systemic chemotherapy using MRI acquired at baseline or early in treatment (1 month follow-up). Tumors with lower ADC at baseline and larger increases in ADC after 1 month of therapy had the greatest response, possibly reflecting initial high cellularity and development of necrosis after therapy, respectively. When the trial completes patient accrual, we will extend the analysis to include the value of combining Ktrans and ADC.

Acknowledgements

No acknowledgement found.

References

1. Konstantinidis IT et al, Ann Surg Oncol 2014; 21:2675

2. Radiologic Society of North America Quantitative Imaging Biomarker Alliance: Dynamic Contrast Enhanced Magnetic Resonance Imaging Technical Committee http://qibawiki.rsna.org/images/7/7b/DCEMRIProfile_v1_6-20111213.pdf.

3. Parker, G.J., et. Al. Magn Reson Med 2006 56(5) p 993-1000.

Figures

Figure 1. Change in sum of diameters (%ΔSD) vs. baseline parameter values. A) %ΔSD after 3 months of treatment vs. baseline mean ADC, B) %ΔSD after 6 months vs. baseline mean ADC, C) %ΔSD after 3 months vs. baseline median Ktrans, D) %ΔSD after 6 months vs. baseline median Ktrans.

Figure 2. %ΔSD vs. change in parameter from baseline to 1 month. A) %ΔSD after 3 months of treatment vs. %ΔADC (BL-1MO) , B) %ΔSD after 6 months vs. %ΔADC (BL-1MO), C) %ΔSD after 3 months vs. %ΔKtrans (BL-1MO), D) %ΔSD after 6 months vs. %ΔKtrans (BL-1MO).

Figure 3. Results of ROC analysis of ability of MRI parameters to predict partial response (PR) after 3 and 6 months of treatment. The table contains the AUC values for each parameter or change in parameter.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
0530