Seeing is believing: A roadmap for implementing bolus-tracked multiphasic CT simulation for ablative radiotherapy of abdominal malignancies

TECHNICAL NOTE

Abdominal malignancies are increasingly treated with ablative radiotherapy, requiring highly accurate target delineation. Unfortunately, many liver and pancreas tumors are nearly invisible in traditional contrast-enhanced CT simulation (CTsim) scans. Targets are frequently contoured based upon their appearance in higher quality diagnostic images, and mapped to planning CT scans via an image registration. Gross tumor volume (GTV) contouring accuracy is limited by the accuracy of this fusion, which is frequently compromised by inconspicuous morphological features, and differences in patient positioning, organ motion, and deformation between the two (or more) image sets – especially if abdominal compression is employed. Reported registration errors for liver [1,2] or pancreas [3] image data are typically on the order of several millimeters, sometimes exceeding 1 cm, with standard deviations of as much as 3-5 mm in each direction [2]; this systematic error is undesirable given the small margins employed in SBRT.

A solution exists: liver and pancreas tumor conspicuity can be substantially enhanced in CTsim data itself, utilizing diagnostic multiphasic contrast enhanced CT (MPCT) protocols. MPCT employs larger contrast volumes injected at high flow rates[4], followed by multiple sequential CT scans, each timed precisely on a patient-specific basis via “bolus tracking” software to capture perfusion and washout differences between tumor and normal tissue. Clear visualization of the tumor in MPCTsim data often eliminates the need for additional diagnostic images, while improving confidence in any remaining registrations.

Despite its efficacy and ease, MPCTsim remains rare outside of its reported deployment at a small number of academic institutions. We believe this is due in part to prior reports of only moderate tumor enhancement in MPCT acquired without patient-specific scan timing or high contrast flow rates (preferably 5 mL/s). We recently adopted an effective bolus-tracked MPCTsim method and published an analysis of its impact on pancreas SBRT target definition[5]. Herein, we describe key details of our experience implementing and performing >100 bolus-tracked MPCT simulation scans over 5 years in a generalized radiation oncology practice (Durham VA Medical Center) for the treatment of primary and metastatic liver and pancreas malignancies, so that others may perhaps apply a similar roadmap at their respective institutions.

MPCT Simulation [Key Details]

Peripheral IV Placement

An 18-gauge IV catheter is preferred; however, if an 18-gauge catheter cannot be inserted, 20-gauge is acceptable provided that the flow rate is reduced. A vein-finder (Veinsite, VueTek Scientific, Gray, Maine) is used to assist IV placement when necessary.

Contrast Injector Programming

We deliver 175 mL of Isovue 300 (Bracco Diagnostics, Monroe Township, NJ) at a flow rate that is selected to match the inserted catheter: 5 mL/s for 18-gauge, or 4 mL/s for 20-gauge, with a safety timeout performed to ensure that the rate does not exceed catheter specifications. A 20 mL saline test bolus precedes, and 40 mL of saline flush is delivered after the contrast injection.

Bolus Tracking and MPCT Timing

Optimal lesion conspicuity in the first two contrast phases is typically brief -- e.g., enhancement persists for approximately 5 s during the late arterial phase -- and thus the scan timing must be carefully controlled. The initial contrast arrival time to abdominal organs is patient specific; this variability can be accounted for by employing bolus tracking software now standard on most CT scanners. Bolus tracking repeatedly measures the mean Hounsfield unit (HU) value in a user-drawn ROI; once a specified HU increase has been detected (the “bolus trigger”), a countdown to initiation of the scan sequence begins. (Note: the trigger typically occurs 15-20 seconds following contrast injection, but we have observed a range of times spanning 10-25 s.)

An ROI is drawn in the descending aorta at the level of the 9^th thoracic vertebrae, and the CT is programmed to acquire 3 sequential scans following a mean ROI rise of 50 HU, according to the following timing:

• late arterial phase (AP) = trigger + 15 s

• portal venous phase (PVP) = trigger + 45 s

• delayed phase (DP) = trigger + 165 s

To ensure that the brief AP lesion enhancement is captured, the AP scan range is set to only include a region surrounding the target lesion itself. According to physician preference, either or both of the PVP or DP scans are configured with a longer scan length appropriate for treatment planning. A fast tube rotation (0.44 s / revolution) is selected for at least the AP and PVP, to minimize the likelihood of missing the fleeting contrast phases.

MPCT Scan and Contrast Sequence

1. Bolus tracking baseline scan, aorta ROI delineation, and threshold selection (baseline + 50 HU)

2. 20 mL saline test injection

3. Contrast injection and bolus tracking initiation

4. Trigger detected

5. AP CT (trigger + 15 s)

6. PVP CT (trigger + 45 s)

7. DP CT (trigger + 165 s)

MPCT Image Evaluation

Correctly timed MPCT scans exhibit the following image characteristics[6,7]:

• AP: Enhancement of hepatic artery and branches, and partial enhancement of the portal vein, but not the hepatic veins. Liver parenchyma enhancement <30% of its maximum. Heterogeneous mottling of the spleen[8].

• PVP: Enhancement of portal and hepatic veins. Maximum liver parenchyma enhancement.

• DP: Enhancement of portal and hepatic veins. Liver parenchyma somewhat enhanced.

Hepatocellular carcinomas (HCC) typically enhance during the AP and washout during the PVP and DP, frequently showing rim enhancement during the PVP and/or DP[6]. Most metastatic lesions and intrahepatic cholangiocarcinomas are best visualized in the PVP and DP, though some exhibit arterial enhancement[7]. Pancreatic cancers [4,5,9] typically appear hypodense in the AP and PVP, but maximum tumor conspicuity may appear in either.

Sample MPCT Simulation Data

Sample MPCT simulation images from an HCC SBRT patient are displayed in Figure 1.

Figure 1.

Breath-hold axial MPCT simulation slices of a liver SBRT patient. The target HCC lesion (solid arrowhead) appears strongly hyperdense in the AP, mildly hyperdense with some rim enhancement in the PVP, and hypodense with conspicuous rim enhancement in the DP scan. In the AP image, the spleen (downward dashed arrow) appears mottled, indicating that the scan was properly timed to capture the brief late AP. The yellow isodose line corresponds to the prescription isodose of 50 Gy.

Discussion

We describe key details necessary to implement MPCT simulation for liver and pancreatic tumors, to reduce systematic GTV contouring errors. We have exported the technique to our academic affiliate where it has been successfully adopted[5]. With this communication, we aim to facilitate more widespread use of a simple, effective method for enhancing tumor visualization critical for ablative radiotherapy.