Quantitative therapy response assessment by volumetric iodine-uptake measurement: Initial experience in patients with advanced hepatocellular carcinoma treated with sorafenib

European Journal of Radiology 82 (2013) 327–334

Contents lists available at SciVerse ScienceDirect

European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad

Quantitative therapy response assessment by volumetric iodine-uptake measurement: Initial experience in patients with advanced hepatocellular carcinoma treated with sorafenib Xu Dai a,b , Heinz-Peter Schlemmer b , Bernhard Schmidt c , Karolin Höh d , Ke Xu a , Tom M. Ganten d , Maria-Katharina Ganten b,∗ a

Department of Radiology, The First Affiliated Hospital of China Medical University, 155 Nanjingbei Street, Shenyang, 110001, China Department of Radiology, Deutsches Krebsforschungszentrum, Im Neuenheimer, Feld 280, 69120 Heidelberg, Germany c Siemens AG, Healthcare Sector, Siemensstr. 1, 91301 Forchheim, Germany d Department of Gastroenterology, Internal Medicine IV, University of Heidelberg, Im, Neuenheimer Feld 410, 69120 Heidelberg, Germany b

a r t i c l e

i n f o

Article history: Received 15 May 2012 Received in revised form 21 October 2012 Accepted 7 November 2012 Keywords: CT density DECT Hepatocellular carcinoma Response evaluation Sorafenib Volumetric iodine-uptake

a b s t r a c t To investigate the volumetric iodine-uptake (VIU) changes by dual-energy CT (DECT) in assessing the response to sorafenib treated hepatocellular carcinoma (HCC) patients, compared with AASLD (American Association for the Study of Liver Diseases) and Choi criteria. Materials and methods: Fifteen patients with HCC receiving sorafenib, monitored with contrast-enhanced DECT scans at baseline and a minimum of one follow-up (8–12 weeks) were retrospectively evaluated. 30 target lesions in total were analyzed for tumor response according to VIU and adapted Choi criteria and compared with the standard AASLD. Results: According to AASLD criteria, 67% target lesions showed disease control: partial response (PR) in 3% and stable disease (SD) in 63%. 33% lesions progressed (PD). Disease control rate presented by VIU (60%) was similar to AASLD (67%) and Choi (63%) (P > 0.05). For disease control group, change in mean VIU was from 149.5 ± 338.3 mg to 108.5 ± 284.1 mg (decreased 19.1 ± 42.9%); and for progressive disease group, change in mean VIU was from 163.7 ± 346.7 mg to 263.9 ± 537.2 mg (increased 230.5 ± 253.1%). Compared to AASLD (PR, 3%), VIU and Choi presented more PR (33% and 30%, respectively) in disease control group (P < 0.05). VIU has moderate consistency with both AASLD (kappa = 0.714; P < 0.005) and Choi (kappa = 0.648; P < 0.005), while VIU showed a better consistency and correlation with AASLD (kappa = 0.714; P < 0.005; r = 0.666, P < 0.005) than Choi with AASLD (kappa = 0.634, P < 0.005; r = 0.102, P = 0.296). Conclusion: VIU measurements by DECT can evaluate the disease control consistent with the current standard AASLD. Measurements are semi-automatic and therefore easy and robust to apply. As VIU reflects vital tumor burden in HCC, it is likely to be an optimal tumor response biomarker in HCC. © 2012 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Hepatocellular carcinoma (HCC) is the third common cause of cancer related death worldwide, and at the time of diagnosis, most patients are in advanced stages [1]. Therapeutic modalities

∗ Corresponding author at: Department of Radiology, Deutsches Krebsforschungszentrum, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany. Tel.: +49 6221 422493; fax: +49 6221 422462. E-mail addresses: [email protected] (X. Dai), [email protected] (H.-P. Schlemmer), [email protected] (B. Schmidt), [email protected] (K. Höh), [email protected] (K. Xu), [email protected] (T.M. Ganten), [email protected] (M.-K. Ganten). 0720-048X/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejrad.2012.11.013

for patients with advanced stage HCC have been limited until the Sorafenib HCC Assessment Randomized Protocol (SHARP) trial was reported [2] in 2008. Sorafenib is a multikinase inhibitor that simultaneously inhibits certain targeted tyrosine kinases, thereby inhibiting tumor proliferation and abrogates neoangiogenesis. It is expected to mainly control and stabilize tumor growth, rather than to shrink lesions [3]. Some recent studies describe early necrosis in hepatic tumors treated with sorafenib [4]. However, since the therapeutic mechanism of sorafenib is different from the conventional chemotherapy, the anatomic response assessment criteria currently in use might be inadequate. In the SHARP study for example, the sorafenib treatment prolonged the clinical progression time apparently compared with the placebo group, while the response rate (CR and PR) was

328

X. Dai et al. / European Journal of Radiology 82 (2013) 327–334

as disappointing as 2% by the assessment criteria of Response Evaluation Criteria in Solid Tumor (RECIST) which are based on size measurement [2]. RECIST relies on the sum of the greatest diameters of target lesions [5]. It is an anatomic size measurement, which is easy to apply in a clinical setting, and is more appropriate to measure cytotoxic effects of chemotherapy. However RECIST fails to show therapy-induced intratumoral changes in response to sorafenib [2,6]. HCC is a heterogeneous and often multifocal tumor, whose growth rates may vary among lesions. And even within the same tumor nodule, there are varying degrees of differentiation, which makes macroscopic growth patterns hard to interpret, underlining the importance of function-based evaluation methods [1,7]. Therefore, both the European Association for the Study of the Liver (EASL) [8] and the American Association for the Study of Liver diseases (AASLD) [9,10] recommended that the assessment of tumor response should incorporate the reduction in viable tumor burden, defined by the area (EASL) or diameter (AASLD) measurement of contrast enhancing tumor on arterial phase imaging to assess the tumor response. Until now, AASLD diameter measurement of viable tumor is widely used in clinic and considered as the standard assessment of treatment efficiency in patients receiving antiangiogenic therapy [11], however it has some limitations as measurement in one single transversal selected slice which may be easily affected by the artificially selected variations, therefore reproducible measurements are very observer dependent [12]. Recently, Choi et al. have proposed the measurement of CT density as a potential indicator of gastrointestinal stromal tumor (GIST) response in patients undergoing targeted therapy [13]. This method had also been applied to other tumor entities, such as the targeted therapy response assessment of HCC [14]. Conversely, a substantial change in tumor density at CT often occurs during targeted therapy. In antiangiogenesis targeted therapy, decreased tumor density of responding area on CT pathologically correlates with the development of tumor necrosis on histopathology, which is derived from the disruption of neoangiogenesis. Tumor necrosis often occurs later than the neovessel disruption and the following blood perfusion reduction. Therefore, decreased CT density changes may not be detected sensitively in the early stage of response. In rare cases, however, HCC response may result in increased density because of the intratumoral hemorrhage, which is a rare effect observed during sorafenib therapy [15], thus result mislead tumor density measurements and failed Choi response assessement. Therefore, ideally, new parameter directly related with neovascularisation should be acquired to assess the density differences related to contrast medium accumulation, or, those related to vital and vascularized tumor tissue. Dynamic contrast-enhanced CT perfusion has been explored as potential new method for assessing response of tumor vascularization to antiangiogenic therapy [6,14]. However in clinic practice the promotion of CT perfusion is prohibited regarding the limitations (e.g. limited coverage of all tumor sites, unstandardized postprocessing, radiation dose) [16]. Dual-energy CT (DECT) is a promising technique used to obtain material specific images. It allows selective quantification and visualization of iodine-related density differences [16,17] and improves the ability to detect contrast agent and to distinguish high-density substances created by iodine from those created by hemorrhage. The iodine map generated by DECT encodes the iodine distribution in each individual CT voxel and subsequently is used to subtract the iodine from the image, which permits selective quantification of iodine-related attenuation (IRA, measured in Hounsfield unit, HU) and volumetric iodine-uptake (VIU, measured in mg) in specified sample tissue based on this material-specific feature [16,17].

Furthermore, as the amount of iodinated contrast medium in tissue depends on its degree of vascularization, the amount of VIU by DECT may be considered as representative of blood perfusion and vascularization in the tumor. One preliminary study [16] had shown that IRA in DECT might be a more robust response parameter than CT density of Choi for it is not influenced by intratumoral hemorrhage. IRA in DECT measured by HU reflects the mean iodine density of the whole volume, it neglects the impact of volume on tumor response assessment. However, the VIU reflects the total amount of iodine in the whole tumor, which is the product of iodine density and tumor volume. We assumed the VIU may reflect the blood perfusion changes in the vital tumor burden more directly and sensitively. First of all proof of principle studies are requested as to demonstrate the feasibility of VIU measurements in solid tumors. Therefore, the purpose of this pilot study was to investigate the VIU changes by DECT in assessing the therapy response in advanced HCC patients treated with sorafenib, compared to the other two vital tumor burden measurements criteria, the current standard AASLD, and Choi. 2. Materials and methods 2.1. Patients demographics We retrospectively analyzed data from all advanced HCC patients who received sorafenib treatment and monitoring by DECT in our institution between September 2010 and January 2012. Baseline contrast enhanced DECT scan was obtained up to one month before sorafenib treatment and at least one follow-up scan in 8–12 weeks under sorafenib treatment. Informed consent was obtained from all patients; the study was approved by our institutional review board. The inclusion and exclusion criterias of patient’s selection were as following: • histologically proven or clinically diagnosed [10] inoperable (due to size, localization etc.) HCC; • hepatic function Child–Pugh Class A or B; • presence of a measurable target lesion showing intratumoral arterial enhancement in contrast enhanced DECT [9]. The measurable diameter should be at least 1 cm and the lesion should be suitable for repeated measurements; • exclusion:prior systemic treatment; possible prior transarterial chemoembolizatoion (TACE), or radiofrequence abladation (RFA) had to be stopped more than 3 months before the initiation of sorafenib therapy; prior treatment with TACE was only permitted if patients had new lesions or apparent recurrence out of the embolized area; In total, 41 HCC patients were examed in our institution with DECT at least one time in this period, including 24 patients treated with sorafinib. At last 15 patients (13 men and 2 women) with 30 measurable target lesions (median 2 lesions/patient, range 1–5) were included in this study, the excluded 9 patients without following up DECT exams. The patients characteristics are displayed in Table 1. 2.2. Sorafenib treatment Patients self-administered sorafenib (Bayer Healthcare, Leverkusen, Germany) orally at a starting dose of 400 mg twice daily and treatment was adapted according to the standard recommendations.

X. Dai et al. / European Journal of Radiology 82 (2013) 327–334 Table 1 Patient demographics. Variable Age (years) Mean Range Gender Males Females Etiology of HCC Alcoholism HBV HCV Cryptogenic Child–Pugh class A B Baseline AFP <20 IU/ml ≥20 IU/ml

Value 67.5 ± 10.2 49–80 Number of patients 13 2 1 2 1 11 6 9 5 10

AFP, ␣-fetoprotein; HBV, hepatitis B virus; HCV, hepatitis C virus.

2.3. DECT scanning Contrast enhanced DECTs, consisting of arterial phase (AP) and portal venous phase (PVP) were performed on the 64-slice dual-energy CT scanner (SOMATOM Definition Flash, Siemens Healthcare Sector, Forchheim, Germany). All patients were scanned in craniocaudal direction from the dome of the liver to the iliac crest during inspiratory breath-hold. Iodinated contrast medium (Iomeprol-400, Iomeron, Bracco Imaging SpA, Milan, Italy) was intravenously administered via an automated dual-syringe power injector (Accutron CT-D, Medtron, Saarbrücken, Germany) according to a body weight (mean 68 ± 15 kg) adapted injection protocol. Mean contrast amount was 97 ± 9 ml, flow rate was 3.9 ± 0.3 ml/s. The timing for the arterial phase scan was determined using the care bolus technique (Siemens Healthcare, Forchheim, Germany), i.e., AP scanning was automatically started 7 s after the attenuation coefficient of abdominal aortic blood reached 120 HU. The PVP images were acquired another 30 seconds after the arterial phase examination. Contrast application was the same for the individual patient in every imaging time point. All AP and PVP images were obtained in the dual energy mode. The DE scan was acquired with a detector collimation of 32 × 0.6 mm; slice 3.0 mm; rotation time, 0.5 s; pitch, 0.6; tube voltage, 100 kV at tube A and 140 kV at tube B. The tube current was adjusted in a real-time manner by the automatic dose modulation protocol Care Dose 4D (Siemens Healthcare, Forchheim, Germany) and the reference effective tube current time product was set 200 mAs for 100 kV tube A and 155mAs for 140 kV tube B. All scans were reconstructed to contiguous axial slices (reconstruction slice thickness, 1.5 mm; increment 1.2 mm in AP and 1.0 mm in PVP) using standard soft tissue reconstruction kernels Siemens D20f. 2.4. Image processing All data were transferred to the workstation using a prototype software syngo.IPIPE (Siemens Healthcare Sector, Forchheim, Germany) To calculate the mean iodine-uptake and density measurements of the tumor in the arterial contrast phase [18] a semi-automatic tumor segmentation algorithm was used. This Dual Energy Liver virtual non-contrast (VNC) algorithm segments a volume of interest after the user has drawn in a rough lesion diameter. Manual refining of the VOI margin was possible if the semi-automated segmentation was not satisfactory in some axial slices. The tumor is

329

segmented by exploiting the local gray-value statistics of the tumor tissue. The contrast medium enhancement is quantified based on a three material decomposition assuming the main components fat, soft tissue and iodine [17]. On the basis of calibration measurements performed by the manufacturer, the algorithm is able to transform spectral information of dual energy data into absolute values of iodine content. Both algorithms (tumor segmentation and LiverVNC) were verified using a dedicated dual energy phantom with iodine inserts and have already been approved by the Food and Drug Administration (FDA) of the USA [16]. Once the segmentation was completed satisfactorily, the software calculated and showed the assessment parameter values automatically, including segmented volume, tumor CT density (measured in HU), and also the VIU (iodine-uptake measured in mg). To minimize potential measurement errors, all target lesions were segmented three times and the mean value of these VOI measurements was used for further analysis. Additionally, the maximum transversal viable tumor diameter (according to AASLD criteria) of each target lesion was measured properly in arterial phase CT images with routine software provided by the manufacturer (SyngoCT Oncology Software, Siemens Healthcare, Forchheim, Germany). Viable tumor was defined as tumor tissue enhancing in arterial phase CE-DECT examinations after contrast medium injection [9,10]. 2.5. Response evaluation by imaging measurement DECT studies were evaluated retrospectively by two radiologists with more than 10 years of diagnostic experiences in a consensus decision. At baseline and follow-up examination target lesions were evaluated with AASLD, Choi and VIU measurement criteria (Table 2). The response evaluation was done per lesion and not per patient; therefore we did not take into consideration changes in non-target lesions, the appearance of new lesions and any change in extrahepatic sites. Disease control is defined as the sum of complete response (CR), partial response (PR) and stable disease (SD). However, the objective response (OR) is defined as the sum of CR and PR, similar to other studies [14]. 2.5.1. AASLD evaluation Unidimensional maximum transversal diameter of viable tumor was measured manually according to AASLD criteria [9,10]. Progressive disease (PD) refers to 20% increase, partial response (PR) to 30% decrease in diameter, complete response (CR) refers to complete disappearance of measurable lesions and stable disease (SD) defined as that between PD and PR. 2.5.2. Choi evaluation Tumor CT density was assessed using CT attenuation coefficients as described by Choi and colleagues in GIST [13]. The Choi criteria have been slightly modified taking into account not the density changes of a ROI (two dimensional region of interest) but the VOI of the tumor (three dimensional volume of interest). PR according to Choi criteria was defined as a decrease in tumor density (measured in Hounsfield units, HU) of 15% or more on CT or a decrease in target size (largest diameter of target lesion, mm) of 10% or more. 2.5.3. VIU evaluation The amount of VIU was also evaluated in segmented VOI of the tumor. Because there is no similar study in iodine-uptake before, analog to Choi criteria and referred with the [18 F]-FDG tumor PR response criteria made by EORTC (European Organization for Research and Treatment of Cancer) positron emission tomography

330

X. Dai et al. / European Journal of Radiology 82 (2013) 327–334

Table 2 Evaluation criteria used in the assessment of target lesion response. AASLD

Choi

VIU

CR PR

No viable lesions ≥30% decrease of viable diameter (vs. baseline)

PD

≥20% increase of viable diameter (vs. baseline) or new lesions

No viable lesions ≥10% decrease in sizea (vs. baseline) or ≥15% decrease in tumor density (HU) ≥10% increase in sizea (vs. baseline) and does not meet criteria of PR by tumor density (HU)

SD

Neither sufficient decrease for PR nor sufficient increase for PD

Neither sufficient decrease for PR nor sufficient increase for PD

No viable lesions ≥10% decrease in sizea (vs. baseline) or ≥15% decrease in target iodine-uptake (mg) ≥10% increase in sizea (vs. baseline) and does not meet criteria of PR by target iodine-uptake (mg) Neither sufficient decrease for PR nor sufficient increase for PD

a

Size is measured by the largest diameter of target lesion.

Table 3

3a Quantitative response assessment parameters at baseline and follow-up Evaluation Criteria

Baseline

Follow-up

Variation (%)

P value

AASLD (diameter, mm) Choi (CT density, HU) VIU (iodine amount, mg)

43.1 ± 30.6 66.5 ± 23.7 155.2 ± 335.8

47.5 ± 35.5 64.2 ± 23.4 170.6 ± 403.5

11.2 ± 21.7 0.6 ± 26.9 80.7 ± 202.1

<0.005 0.004 0.015

3b Quantitative response assessment parameters at baseline and follow-up of VIU responder and non-responder group VIU

Number

VIU-baseline

VIU-followup

Variation (%)

Responder Non-responder Total

18 12 30

149.5 ± 338.3 163.7 ± 346.7 155.2 ± 335.8

108.5 ± 284.13 263.9 ± 537.2 170.6 ± 403.5

−19.1 ± 42.9 230.5 ± 253.1 80.7 ± 202.1

Note: Values are mean ± standard deviation. Responder means disease control group, including PR and SD; Non-responder means disease progressive group, which is PD.

(PET) study group [13,19] in GIST, PR of VIU was also defined as a decrease in iodine-uptake (measured in mg) of 15% or more on CT or a decrease in target size (largest diameter of target lesion, cm) of 10% or more.

2.6. Statistical analysis The values for continuous variables are expressed as mean ± standard deviation, and categorical variables as the percentage frequencies and percentages. The categorical variables comparisons between groups were done using the 2 test and the consistency evaluation with kappa statistical analysis. The continuous variables comparisons in groups were done using the one sample t-test and the correlations between different groups using Spearman correlation analysis. Significance is defined by P < 0.05. All statistical analysis were performed with the SPSS software package (version 17.0; SPSS, Inc., Chicago, Ill).

3. Results

3.2. Comparison of response evaluation results The sum of response evaluation results of these 30 target lesions according to AASLD, Choi and VIU criteria are shown in Table 4. According to AASLD criteria, 20 (66%) of 30 target lesions showed disease control (PR+SD, no CR in our group), including PR in 1 (3%) and SD in 19 (63%), and the remaining 10 progressed (33%). Choi and VIU measurements identified disease control in similar percentage to AASLD criteria, which was 19 (63%) with Choi and 18 (60%) with VIU, respectively (P > 0.05) (Table 4, Fig. 1). Compared with AASLD, there were no apparent differences between Choi and VIU criteria in classifying the disease control and progression response after sorafenib therapy. Concerning the OR (PR, no CR in our group), there are significant differences among these three imaging assessment methods. Compared with AASLD, most response categories of Choi and VIU shifted from SD toward PR (Table 4). AASLD labeled 1 (3%) of 30 target lesions with PR, in which 9 (30%) with Choi and 10 (33%) with VIU (P < 0.05, compared with AASLD respectively). Choi and VIU presented a significantly higher PR ratio than AASLD in disease control group (example for PR-response in VIU shown in Fig. 2).

3.1. Quantitative response assessment parameters

3.3. Consistency analysis of three imaging measurements

The baseline and follow-up parameter changes are shown in Table 3a and 3b. There was an average 11.15% ± 21.73% change in AASLD measurement (P < 0.005); slight change in mean CT density from 66.51 ± 23.74 HU to 64.19 ± 23.40 HU (variation percentage, 0.62% ± 26.94%) (P = 0.004) and major change in mean VIU from 155.16 ± 335.76 mg to 170.63 ± 403.45 mg (variation percentage, 80.73% ± 202.09%) (P = 0.015). For disease control group (PR and SD), change in mean VIU was from 149.5 ± 338.3 mg to 108.5 ± 284.1 mg (variation decreased 19.1 ± 42.9%); and for progressive disease group (PD), change in mean VIU was from 163.7 ± 346.7 mg to 263.9 ± 537.2 mg (variation increased 230.5 ± 253.1%). VIU had a significantly higher variation percentage value than Choi or AASLD.

The evaluation results of 30 target lesions by VIU versus AASLD were shown in Table 5a; VIU versus Choi in Table 5b; Choi versus AASLD in Table 5c. The results evaluated by three imaging measurements had significant differences between each other (P < 0.01).

Table 4 Comparison of response evaluation results. Evaluation Criteria

CR

AASLD Choi VIU

0 0 0

PR 1/30 (3%) 9/30 (30%) 10/30 (33%)

SD 19/30 (63%) 10/30 (33%) 8/30 (27%)

PD 10/30 (33%) 11/30 (37%) 12/30 (40%)

X. Dai et al. / European Journal of Radiology 82 (2013) 327–334

331

Fig. 1. Baseline CT (A, B, C) and follow-up CT (D, E, F) of a patient with mass type of HCC after 6 weeks of sorafenib treatment. Arterial phase (A, D), Virtual 120-kV (B, E) and corresponding iodine maps (C, F). Target lesion was segmented with color encompassed area in transversal CT slices as shown in B, E by semi-automatic software. This case demonstrated a good consistency of SD response in AASLD (viable diameter decrease 2.69%; baseline, 122.5 mm, follow-up, 119.2 mm), Choi (CT density increase 0.53%; baseline, 56.7 HU, follow-up, 57 HU) and VIU (iodine-uptake decrease 8.09%, baseline 1325.5 mg, follow-up, 1218.3 mg) assessments. From the beginning of sorafenib treatment, this patient had been still alive until the end of the study survey, survived more than 16 months now.

Fig. 2. Baseline CT (A, B, C) and follow-up CT (D, E, F) of a patient with mass type of HCC after 3 months of sorafenib treatment. Arterial phase (A, D), Virtual 120-kV (B, E) and corresponding iodine maps (C, F). Target lesion was segmented with color encompassed area in transversal CT slices as shown in B, E by semi-automatic software. This case demonstrated a good consistency of SD response in AASLD (viable diameter decrease 1.68%; baseline, 41.6 mm, follow-up, 40.9 mm) and Choi (CT density decrease 7.88%; baseline, 44.4 HU, follow-up, 40.9 HU); while VIU demonstrated a PR response (iodine-uptake decrease 48.87%, baseline 44.4 mg, follow-up, 22.7 mg). From the beginning of sorafenib treatment, this patient had been survived 32 months which was more than the mean survival time of this group (Mean survival 13.2 ± 9.73 m) and died of HCC progression at last.

When regarding the disease control, varying degrees of consistency among three imaging measurements were demonstrated by kappa analysis, as shown in Table 6. VIU has better consistency with AASLD than others (Fig. 3). Spearman correlation demonstrated a good correlation of r = 0.666 (P < 0.005) between VIU and the AASLD criteria for all 30 target lesions, which is better than the correlation between Choi and AASLD (r = 0.102, P = 0.296), also VIU and Choi (r = 0.466, P = 0.005) (Table 6). VIU correlated better with AASLD than Choi.

4. Discussion Response monitoring of targeted therapies is currently one major challenge in imaging. Because sorafenib treatment in advanced HCC can result in initially tumor size unchanged but altered tissue composition. The standard AASLD criteria focus on diameter measurements of viable tumor parts but are manually performed and therefore less objective. Ongoing developments in targeted therapies increase the need of a practicable objective

332

X. Dai et al. / European Journal of Radiology 82 (2013) 327–334

Table 5

5a Comparison of response evaluation results: VIU vs. AASLD AASLD, n (%)

VIU, n (%)

PR SD PD

PR

SD

PD

1 (3) – –

8 (27) 8 (27) 3 (10)

1 (3) – 9 (30)

5b Comparison of response evaluation results: VIU vs. Choi Choi, n (%)

VIU, n (%)

PR SD PD

4.1. Comparison of response evaluation results

PR

SD

PD

6 (20) 0 3 (10)

3 (10) 7 (23) 0

1 (3) 1 (3) 9 (30)

5c Comparison of response evaluation results: Choi vs. AASLD AASLD, n (%) PR Choi, n (%)

1 (3) – –

PR SD PD

reproducible biomarker for monitoring tumor response. VIU is a new assessment parameter along with the development of DECT and assumed to reflect the vital tumor burden by the iodine-uptake of viable tumor. In contrast enhanced DECT, the iodinated contrast medium in lesion is mainly brought by arterial blood perfusion of viable tumor. Under the relatively fixed concentration and flow rate of iodinated contrast medium, the VIU should be proportional to the blood perfusion in segmented lesions. Reflecting biologic information of the tumor, VIU could be a promising tool for evaluating new targeted therapies in HCC.

SD

PD

6 (20) 10 (33) 3 (10)

2 (7) – 8 (27)

The assessment of intratumoral vital tumor burden still is difficult in HCC under targeted therapy. AASLD criteria are now accepted as the widely used assessment method in therapy of advanced HCC [11]. Therefore we selected the AASLD criteria as the gold standard for evaluation. Furthermore, VIU measurements were compared with Choi and AASLD to see the consistency among these three imaging response criteria in targeted therapy. 4.1.1. VIU compared with standard AASLD This study shows that VIU assessments had good consistency in the evaluation of disease control (PR + SD) and progression (PD) with AASLD criteria. While regarding the disease control group, VIU presented significantly higher PR ratio than AASLD.

Fig. 3. Baseline CT (A, B, C) and follow-up CT (D, E, F) after 3 months of sorafenib treatment. Arterial phase (A, D), Virtual 120-kV (B, E) and corresponding iodine maps (C, F). Target lesion was segmented with color encompassed area in transversal CT slices as shown in B, E by semi-automatic software referring to PR according to Choi (CT density decrease 36.74%; baseline, 55.8 HU, follow-up, 35.3 HU) after 3 months of sorafenib therapy, while the VIU (iodine-uptake increase 65.22%; baseline, 1081.8 mg, follow-up, 1787.3 mg) and AASLD (viable diameter increase 62.22%; baseline, 99.8 mm, follow-up, 161.9 mm) demonstrated a PD response. This patient died of HCC progression and survivd only 3 months from the beginning of sorafenib therapy which confirmed the PD response. This example shows VIU having better consistency with AASLD than Choi in some patients.

Table 6 Consistency and correlation analysis. Evaluation criteria

Kappa

VIU vs. AASLD VIU vs. Choi Choi vs. AASLD

0.714 0.648 0.634

a

CI: Confidence interval.

95% CIa level

P

lower

upper

0.457 0.368 0.344

0.972 0.928 0.924

<0.005 <0.005 <0.005

Spearman correlation r

P

0.666 0.466 0.102

<0.005 0.005 0.296

X. Dai et al. / European Journal of Radiology 82 (2013) 327–334

In our VIU assessment criteria, PR of VIU amount is defined as a decrease in iodine-uptake of 15% or more on DECT. The amount of VIU is directly affected by the perfusion of iodinated contrast medium in the vital tumor, so it could reflect intratumoral blood perfusion or vascular changes caused by antiangiogenic effects of sorafenib more directly and sensitively. Many target treated HCC studies have observed the limited changes in tumor size, but with decrease of blood perfusion due to antiangiogenic effects by MRI or CT perfusion studies [2,4,6]. Our VIU results by DECT also showed the intratumoral vascularization changes caused by sorafenib, however these were sometimes not associated with significant viable tumor shrinkage thus resulting in more SD responses according to AASLD. At the same time, the threshold of target size decreases dropped from 30% to 10% variation as PR classification in VIU measurements seemed to make them more sensitive to classify PR than AASLD. 4.1.2. VIU compared with Choi In our retrospective study, VIU correlates moderately with Choi criteria and there are no apparent differences between VIU and Choi in evaluating the lesions of PD. This is in line with a previous study in GIST tumors [16], demonstrating good correlation between IRA and Choi in the response evaluation. Whereas in this study VIU has better consistency and correlation with AASLD than Choi. Both VIU and Choi can reflect the iodinated contrast medium accumulation in the vital parts of tumor thus can be used to evaluate the intratumoral blood and vascular changes. However, the mean tumor density of Choi reflects not only the accumulation of iodine, but also the mixed internal materials of tumor, as vital tumor, necrosis tissue even hemorrhage caused by targeted therapies. Intratumoral hemorrhage can raise the tumor density and misleads the evaluation of therapy response. VIU is only related with the accumulation of contrast medium, which is similar to the measurement of contrast enhanced vital tumor tissue by AASLD. This is maybe one of the reasons VIU is correlating better than CT density with AASLD in disease control evaluations. Furthermore, VIU had significantly higher percentage variation values than Choi, which should reflect the vital and vascularized tumor tissue changes more sensitively in the targeted therapy response monitoring. VIU of DECT is assumed to reflect perfusion in vital tumor, which makes VIU a promising imaging biomarker reflecting vital tumor burden more precise than current standards. Although the number of patients in this study was limited, these data also strongly suggest that quantified VIU by DECT may be assumed as a more sensitive and promising response predictor than CT density in targeted therapy of HCC. This potential should be evaluated in future multi center and larger studies, correlating VIU measurements with clinical outcome. 4.2. Practicability of quantitative VIU measurements in HCC therapy response monitoring With DECT and semi-automatic post-processing technique, VIU measurements (iodine-uptake measured in mg) as a biomarker for tumor response monitoring are easily and robust to perform. AASLD criteria, as the current standard of tumor response monitoring in HCC, demands manual transversal slice selections and unidimensional measurements of maximum diameter of the contrast enhanced part of tumor. As HCC is a heterogeneous and often multifocal tumor, these manual selections and measurements are difficult and there is some inter- and intraobserver variation of results, which may lead to different response categorization in follow-up studies [12]. Theoretically, the VIU measurement may be able to provide more accurate and robust blood perfusion and vascularization information based on the overall volumetric

333

distribution of iodinated contrast medium, regardless of the heterogeneous necrosis and volume changes caused by targeted therapy or tumor progression. As the software provides besides Volume Iodine Uptake simultaneously the segmented tumor volume, maximum diameter, and tumor CT density, accordingly without an additional native scan assessment of functional response criteria can be performed supplementary to RECIST in one go. Since dual energy scanner of the new generation using tin-filter technology allow scanning protocols with equivalent radiation dose to single energy scans, this multiparametric therapy monitoring approach is feasible at decreased radiation dose compared to multiphasic CT [16,20]. The reproducibility of target segmentation is still an important point in the proper VIU response evaluation of HCC therapy. To qualify this point, we checked all the segmentations of 30 target lesions and necessary manual corrections were done next to the semi-automatic procedure in some segmentations with unsatisfactory margins, especially in large, irregular and heterogeneous lesions. Future advances in software may improve the qualified semi-automatic segmentation process better, along with the deep understandings of VIU in HCC response monitoring. 4.3. Limitations of this study First the patient sample size for this pilot study is relatively small. For the main aim to compare the specific quantified VIU and density, only responses of target lesions were evaluated and the overall responses of patients were not done for the small patient sample. Larger patient sampling and overall response of patient should be done in future studies. Second, the retrospective character of the study. Further prospective follow-up studies should be done to verify whether VIU measurement is truly better correlated with clinical efficacy in HCC targeted therapy in larger clinical trials. Additionally, the VIU assessment criteria used in our study was derived from the reference of the [18 F]-FDG PET response evaluation criteria and the Choi criteria in GIST. In a second step, further quantitative studies should correlate VIU with other independent tumor response parameters, such as overall survival, to define iodine-uptake thresholds more appropriate for the different response categories, and to evaluate the value of VIU as a more independent tumor response parameter in the monitoring of HCC targeted therapy. 5. Conclusion In this pilot study we could show that VIU measurements by DECT can evaluate the disease control consistent with the current standard criteria of AASLD. VIU is likely to be an optimized tumor response biomarker, first reflecting vital tumor burden in HCC, second, measurements being semiautomatic and therefore easy and robust to apply and less observer dependent in general. Future studies, correlating VIU measurements with clinical outcome will clarify its value in predicting tumor response. Conflict of interest None of the authors have identified a conflict of interest. References [1] El-Serag HB, Rudolph KL. Hepatocellular carcinoma: epidemiology and molecular carcinogenesis. Gastroenterology 2007;132(7):2557–76. [2] Llovet JM, Ricci S, Mazzaferro V, et al. Sorafenib in advanced hepatocellular carcinoma. New England Journal of Medicine 2008;359(4):378–90. [3] Wilhelm SM, Adnane L, Newell P, Villanueva A, Llovet JM, Lynch M. Preclinical overview of sorafenib, a multikinase inhibitor that targets both Raf and VEGF

334

[4]

[5]

[6]

[7]

[8]

[9] [10]

[11]

[12]

X. Dai et al. / European Journal of Radiology 82 (2013) 327–334 and PDGF receptor tyrosine kinase signaling. Molecular Cancer Therapeutics 2008;7(10):3129–40. Horger M, Lauer UM, Schraml C, et al. Early MRI response monitoring of patients with advanced hepatocellular carcinoma under treatment with the multikinase inhibitor sorafenib. BMC Cancer 2009;9:208–18. Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). European Journal of Cancer 2009;45(2):228–47. Jiang T, Kambadakone A, Kulkarni NM, Zhu AX, Sahani DV. Monitoring response to antiangiogenic treatment and predicting outcomes in advanced hepatocellular carcinoma using image biomarkers, CT perfusion, tumor density, and tumor size (RECIST). Investigative Radiology 2012;47(1):11–7. Efremidis SC, Hytiroglou P, Matsui O. Enhancement patterns and signal-intensity characteristics of small hepatocellular carcinoma in cirrhosis: pathologic basis and diagnostic challenges. European Radiology 2007;17(11):2969–82. Bruix J, Sherman M, Llovet JM, et al. Clinical management of hepatocellular carcinoma. Conclusions of the Barcelona-2000 EASL conference. European Association for the Study of the Liver. Journal of Hepatology 2001;35(3):421–30. Lencioni R, Llovet JM. Modified RECIST (mRECIST) assessment for hepatocellular carcinoma. Seminars in Liver Disease 2010;30(1):52–60. Bruix J, Sherman M. American Association for the Study of Liver Diseases. Management of hepatocellular carcinoma: an update. Hepatology 2011;53(3):1020–2. Spira D, Fenchel M, Lauer UM, et al. Comparison of different tumor response criteria in patients with hepatocellular carcinoma after systemic therapy with the multikinase inhibitor sorafenib. Academic Radiology 2011;18(1):89–96. van Kessel CS, van Leeuwen MS, Witteveen PO, Kwee TC, Verkooijen HM, van Hillegersberg R. Semi-automatic software increases CT measurement accuracy but not response classification of colorectal liver

[13]

[14]

[15]

[16]

[17] [18]

[19]

[20]

metastases after chemotherapy. European Journal of Radiology 2012;(January), http://dx.doi.org/10.1016/j.ejrad.2011.12.026. Choi H, Charnsangavej C, Faria SC, et al. Correlation of computed tomography and positron emission tomography in patients with metastatic gastrointestinal stromal tumor treated at a single institution with imatinib mesylate: proposal of new computed tomography response criteria. Journal of Clinical Oncology 2007;25(13):1753–9. Faivre S, Zappa M, Vilgrain V, et al. Changes in tumor density in patients with advanced hepatocellular carcinoma treated with sunitinib. Clinical Cancer Research 2011;17(13):4504–12. Rombolà F, Caravetta A, Mollo F, Spinoso A, Peluso L, Guarino R. Sorafenib, risk of bleeding and spontaneous rupture of hepatocellular carcinoma. A clinical case. Acta Medica (Hradec Kralove) 2011;54(4):177–9. Apfaltrer P, Meyer M, Meier C, et al. Contrast-enhanced dual-energy CT of gastrointestinal stromal tumors: is iodine-related attenuation a potential indicator of tumor response? Investigative Radiology 2012;47(1): 65–70. Johnson TR, Krauss B, Sedlmair M, et al. Material differentiation by dual energy CT: initial experience. European Radiology 2007;17(6):1510–7. Llovet JM, Di Bisceglie AM, Bruix J, et al. Design and endpoints of clinical trials in hepatocellular carcinoma. Journal of the National Cancer Institute 2008;100(10):698–711. Young H, Baum R, Cremerius U, et al. Measurement of clinical and subclinical tumour response using [18 F]-fluorodeoxyglucose and positron emission tomography: review and 1999 EORTC recommendations. European Organization for Research and Treatment of Cancer (EORTC) PET Study Group. European Journal of Cancer 1999;35(13):1773–82. Mangold S, Thomas C, Fenchel M, et al. Virtual nonenhanced dual-energy CT urography with tin-filter technology: determinants of detection of urinary calculi in the renal collecting system. Radiology 2012;264(1):119–25.