British Journal of Radiology (2005) 78, 105-109
© 2005 British Institute of Radiology
doi: 10.1259/bjr/13525061
Comparison of the hepatic perfusion index measured with gadolinium-enhanced volumetric MRI in controls and in patients with colorectal cancer
J J Totman, DCR(R), MSc
1
R L O'Gorman, MSc
2,3
P A Kane, MBBS, MRCP, FRCR
1 and
J B Karani, BSc, MBBS, FRCR
1
1 Departments of Radiology, 2 Neuroimaging and 3 Medical Engineering and Physics, King's College Hospital, Denmark Hill, London SE5 9RS, UK
 |
Abstract
|
|---|
The aim of the study was to adapt the methodology established for dynamic CT measurements of the hepatic perfusion index (HPI) to MRI, and to assess the potential role of MRI measurements of the HPI in detecting regional alterations in liver perfusion between patients with colorectal liver metastases and normal controls. The HPI was evaluated from serial T1 volume acquisitions acquired over the course of a Gd-DTPA bolus injection. Time-course data from regions of interest in the liver, spleen and aorta were used to calculate the HPI; and HPI data from control subjects were compared with data from patients with known colorectal metastases. Significant differences were found between the relative portal perfusion and hepatic perfusion indices calculated for the patient and control groups (p<0.005). These results suggest that hepatic perfusion indices can be derived using MRI-based methods, and that these perfusion indices are sensitive to differences in liver perfusion associated with established metastatic liver disease on imaging. This technique may contribute to the early detection of liver metastases, allowing early surgical intervention and improved patient survival.
|
Introduction
|
|---|
Colorectal cancer is the second most common cause of death from cancer in the UK. Currently 33 000 new cases are diagnosed each year and the incidence is increasing yearly by 1% in men [1]. For patients with colorectal malignancy, the presence of liver metastases is the most accurate predictor of survival. Following apparently curative surgical resection of the primary tumour, 50% of patients will die within 5 years, the majority from disseminated disease. Hepatic involvement is a feature in 50% of patients, and in almost one-third of cases the liver may be the only site of metastatic disease found at autopsy [2]. Improved survival following liver resection is now achievable with 5-year disease-free survival rates of 20–45% and up to 60% in unifocal disease [3]. Without surgery, the 5-year survival rate for these patients is effectively zero [4].
Detection of focal liver lesions pre-operatively is dependent on fundamental differences in the tissue characteristics of tumour and normal liver: first, an intrinsic difference in tissue density and cellular components; second, differences in vascularity and tissue haemodynamics. Current imaging techniques can reliably demonstrate metastases of 1 cm or larger using superparamagnetic iron oxide (SPIO)-enhanced MRI and multidetector CT. The sensitivity of imaging methods for detection of lesions smaller than 1 cm is in the region of 50% when surgery and intraoperative ultrasound are used as the gold standard, but the accuracy of this reference standard itself cannot be established [5]. Even if detected, the characterization of lesions of 1 cm or less may be suboptimal on pre-operative imaging. It is these undetected, subcentimetre lesions that typically account for early disease recurrence within the liver and failure of surgical cure [5]. It is now established practice in centres carrying out liver resection to use a focused, multimodality approach, using CT, MRI and PET in order to enhance detection of lesions, and intrahepatic and extrahepatic staging.
One approach proposed to detect occult metastases is based on the alteration in liver blood flow that develops with metastatic seeding in the liver. The ratio of hepatic arterial to total liver blood flow (hepatic perfusion index, HPI), was first investigated using dynamic scintigraphy, and found to be abnormal in 94% of patients with colorectal liver metastases [6]. Furthermore, of those patients who developed liver metastases within 3 years of their original primary resection, 87% had an abnormal HPI at presentation [7]. The methodology developed for measurement of the HPI with scintigraphy was adapted for use with dynamic CT, with similar results [8, 9]. Doppler ultrasound has also been used to identify flow changes in the afferent vessels, giving rise to a Doppler perfusion index (DPI) [10]. All of the above work has been, by necessity, retrospective, but prospective animal studies by Cuenod et al [11] and Nott et al [12] have indicated a high degree of sensitivity to occult metastases for this technique. The aim of the present study was to adapt the methodology established for dynamic CT measurements of the HPI to MRI, and to assess the potential role of MRI measurements of the HPI in the detection of changes in liver perfusion between patients with colorectal liver metastases and controls.
|
Materials and methods
|
|---|
The patient group consisted of eight men and four women (mean age 65 years, range 44–78 years), with colorectal primary confirmed on biopsy or resection, and liver metastases evident on imaging. The control group consisted of eight men and six women (mean age 45 years, range 21–70 years), taken from patients referred for routine contrast-enhanced spine imaging, with no history of neoplastic disease. This study was performed with the approval of the local research ethics committee, and informed consent was obtained from all subjects prior to MRI. Imaging was performed with a Siemens 1T Harmony scanner (Siemens, Erlangen, Germany) using a modified volume-interpolated breath-hold examination sequence (VIBE) [13]. A total scan time of 1:15 min covers the necessary extent of the perfusion curves, with a maximum sampling rate of 3 s, which is within the constraints recommended by Henderson et al [14] for the accurate sampling of time course data. The sequence was acquired while the patient breathed gently in order to avoid blood flow variations introduced by changes in intra-abdominal pressure [15]. This can introduce periodic motion-induced signal changes, which although present can be readily identified as artefact. To allow for future parametric analysis, the entire volume of the liver was examined, resulting in 25 volumes of 36, 5 mm thick partitions. The spatial resolution achievable within the constraints imposed by the prescribed temporal resolution and liver size results in a matrix size of 128 x 67, covering a field of view of 400 mm x 300 mm. A repetition time of 2 ms, an echo time of 0.83 ms, and a bandwidth of 1150 Hz were employed to enable fast sampling. Typically a flip angle of 12° is used with the VIBE sequence as this produces maximum sensitivity to contrast enhancement. However, phantom studies investigating the VIBE signal dependence on flip angle and gadolinium concentration demonstrated that a flip angle of 12° yields maximum sensitivity to any variation in flip angle. Since higher flip angles demonstrated a much reduced sensitivity to changes in flip angle whilst retaining a linear relationship with contrast enhancement, a flip angle of 90° was selected in order to reduce the impact of B1 and B0 field variations.
The patients were cannulated with a 22-gauge venflon in an antecubital vein, which was flushed with saline and connected to a MRI-compatible injection kit (Medex, Annecy-Le-Vieux, France). The bolus injection of the manufacturer's recommended dose of 0.2 ml kg–1 Gd-DTPA (Magnavist, Schering Germany), typically 16 ml, was manually administered and timed to produce an injection rate of 4 ml s–1 at the start of the perfusion sequence; and followed immediately by a 20 ml saline flush, similarly timed. The injection duration was in the order of 4 s, plus an additional 5 s for a saline flush. Overall the total examination time was only slightly increased from the examination times associated with conventional contrast-enhanced imaging, since the necessary increase in preparation time is counterbalanced by the need to bring the patient out of the scanner for their contrast injection.
A single central slice was selected where both the liver and spleen were visible, and time intensity curves were derived from regions of interest (ROI) drawn manually by a single clinical specialist Radiographer, and then checked by an experienced Consultant Hepatobiliary Radiologist in the aorta, liver and spleen (Figure 1)
. The ROIs in the liver were drawn to encompass the parenchyma and metastases, if present, but no major vessels. The HPI was evaluated using a semi-automated program written locally in MATLAB (MathWorks Inc., Natick, MA), following the methodology established for CT by Blomley et al [8]. With this approach, the relative hepatic perfusion is calculated by dividing the maximal slope of the liver curve prior to the time of peak splenic attenuation by the maximal aortic enhancement. The relative portal perfusion is calculated based on the assumption that the liver and spleen receive a similar arterial bolus. Using this model, the hepatic arterial contribution is removed from the liver curve by subtracting a scaled version of the spleen curve (scaled by the ratio of hepatic/splenic perfusion), to yield a curve approximating to the portal contribution to the liver perfusion. The relative portal perfusion is then calculated as the peak up-slope of this curve divided by the peak rise in the splenic or portal vein signal, and the HPI is calculated as the ratio of the apparent hepatic arterial perfusion to the total liver perfusion [8]. In total the data processing requires approximately 15 min per patient, since at present the time-course data must be entered manually.
View larger version (86K):
[in this window]
[in a new window]
|
Figure 1. Axial section from one of the colorectal cancer patients showing extracted time-course data for regions of interest in the liver, spleen and aorta. The liver curve demonstrates significant early enhancement prior to the peak of the spleen curve, consistent with an increase in the hepatic component of perfusion and a corresponding decrease in the portal component.
|
|
The statistical analysis was performed after importing the perfusion data into SPSS version 10.1 for Windows (SPSS Inc., Chicago, IL). Levene's test was used to assess the equality of variances between the patient and control groups, and differences in relative hepatic perfusion, relative portal perfusion, and HPI between the two groups were evaluated with an unpaired 2-tailed t-test.
|
Results
|
|---|
Significant differences were seen between the relative portal perfusion and hepatic perfusion indices calculated for the patient and control groups (p<0.005), with patients demonstrating reduced portal perfusion and increased HPI relative to controls. These results are in good agreement with data reported from previous CT and Doppler ultrasound studies [8–10, 16]. In addition, the patients demonstrated increased hepatic perfusion relative to the controls, although this trend did not reach significance (p=0.063). The distribution of calculated perfusion indices is shown in Figure 2, and the distribution of relative portal perfusion values is given in Figure 3. (Boxplots show median and interquartile ranges.)
View larger version (8K):
[in this window]
[in a new window]
|
Figure 2. Median and interquartile ranges for the hepatic perfusion index (HPI) measured in patients with colorectal metastases and control subjects (p<0.005, unpaired t-test).
|
|
View larger version (9K):
[in this window]
[in a new window]
|
Figure 3. Median and interquartile ranges for the relative portal perfusion measured in patients with colorectal metastases and control subjects (p<0.005, unpaired t-test).
|
|
|
Discussion
|
|---|
Dynamic scintigraphy, CT and Doppler ultrasound have been used to develop equivalent marker indices, though all three techniques have recognized drawbacks. Both CT and scintigraphy involve exposure to ionizing radiation and are "stand alone" procedures, which cannot be easily incorporated into routine diagnostic scanning. In CT, iodinated contrast administration has the attendant risk of anaphylaxis and to date, HPI has only been demonstrated on a single slice, although advances in CT may now allow improved coverage. Doppler ultrasound DPI data is obtained during an extended ultrasound examination, but is operator-dependent, with significant interobserver and intraobserver variability [17]. Furthermore, DPI does not provide any information regarding the anatomical distribution of occult lesions, and relies upon "normal" hepatic arterial anatomy, but only 70% of the population exhibit such anatomy [17].
An MRI-based approach has several perceived advantages over CT and the equivalent marker indices developed for scintigraphy and Doppler ultrasound, including the lack of ionizing radiation and operator dependence. Gadolinium DTPA is also considered an inherently safer contrast agent than those used in CT, and the gadolinium volume used is small, allowing a more compact bolus to be achieved, which aids data processing. The technique can be used with abnormal hepatic anatomy, and can assess the whole liver volume rather than a single slice. This MR-based technique also offers increased spatial and temporal resolution (particularly in comparison with ultrasound and radionuclide based methods), and can be performed in conjunction with high-resolution structural MRI, enabling a more extensive evaluation of tissue morphology and function to be performed.
The linear relationship between concentration of contrast and attenuation of signal in CT allows the perfusion values obtained from CT measurements by this method to be expressed in ml min–1 ml–1. With MRI, the relationship between signal intensity and contrast concentration is non-linear and so this type of quantification is much more complex [18]. Nonetheless, as a ratio, the HPI can be calculated from MRI data, and in this preliminary study appears to represent a valid physiological indicator of abnormal perfusion, yielding results consistent with findings reported from previous dynamic CT, Doppler ultrasound and scintigraphy studies [8–10, 16]. However our preliminary observations have not determined the potential modulation of perfusion by other liver tumours, either benign or malignant, or the effects of pre-existing liver disease or chemotherapy.
Follow-up imaging studies will be required in order to assess the temporal consistency and intrasubject variability of the HPI measurements. However, this technique can potentially enable the early identification of patients with multiple areas of abnormal perfusion but no correlative anatomical lesions, likely to demonstrate early recurrence. Analysis of ROIs in various slices and positions throughout the entire liver volume demonstrate comparable results with those observed on the central slice, but fully automated parametric analysis of the HPI MRI data could offer further information regarding regional variations in perfusion throughout the liver. Intuitively, if these regional perfusion changes can be mapped onto the standard axial images, then the multiplanar visualization of these will allow segmental localization within the liver. These patients can then be managed more effectively with increased surveillance, in order to facilitate early detection on anatomical imaging and avoid porto-portal spread. This will need to be examined in a long-term follow-up study, with a large cohort of patients, following excision of their primary tumour. Conversely, a normal perfusion study would allow an extended interval of surveillance, and provide a further marker of continued remission in long-term follow-up. Such a pathway would be potentially more reproducible than ultrasound and avoid the cumulative radiation risk of CT. In addition, it would improve efficacy with resection, cryoablation and chemotherapy with small volume disease, and prevent further dissemination through the portal venous system.
|
Conclusions
|
|---|
Surgical resection of colorectal metastases may be curative in appropriately selected patients, but appropriate selection of these patients is dependent on accurate radiological detection and staging of these tumours pre-operatively. We propose a MRI-based technique for measuring the HPI, and demonstrate that the perfusion indices measured using this technique can distinguish differences in the HPI and portal perfusion associated with metastatic seeding in the liver. This technique has the potential to provide a valuable method for early detection of "occult lesions", prompting early surgical or radiological intervention.
|
Acknowledgments
|
|---|
We express our deep gratitude to the MRI Radiographers at King's College Hospital, and Dr P E Summers PhD (Institute of Neuroradiology University Hospital, Zurich), and Dr Adam Waldman (Department of Radiology, Charing Cross Hospital, London, UK) for their valuable advice regarding this study.
Received for publication March 24, 2004.
Revision received August 6, 2004.
Accepted for publication September 20, 2004.
|
References
|
|---|
- www.cancerscreening.nhs.uk NHS Cancer Screening Programmes, The Manor House, 260 Ecclesall Road South, Sheffield S11 9PS.
- Abrahams HL, Spiro R, Goldstein N. Metastases in carcinoma. Cancer 1950;3:74–85.[CrossRef][Medline]
- McArdle CS, Hole DJ. Outcome following surgery for colorectal cancer: an analysis by hospital after adjustment for case mix and deprivation. Br J Cancer 2002;86:331–5.[CrossRef][Medline]
- Stangl R, Altendorf-Hofman A, Charnley RM, Scheele J. Factors influencing the natural history of colorectal liver metastases. Lancet 1994;343:1405–10.[CrossRef][Medline]
- Ward J, Naik KS, Guthrie JA, Wilson D, Robinson PJ. Hepatic lesion detection: comparison of MR imaging after the administration of superparamagnetic iron oxide with dual-phase CT by using alternative-free response receiver operating characteristic analysis. Radiology 1999;210:459–66.[Abstract/Free Full Text]
- Breedis C, Young G. The blood supply of neoplasms in the liver. Am J Pathol 1954;30:969–85.
- Cooke DA, Parkin A, Wiggins P, Robinson PJ, Giles GR. Hepatic perfusion index and the evolution of liver metastases. Nucl Med Commun 1987;8:970–4.[Medline]
- Blomley MJ, Coulden R, Dawson P, Kormano M, Donlan P, Bufkin C, et al. Liver perfusion studied with ultrafast CT. J Comput Assist Tomogr 1995;19:424–33.[Medline]
- Miles KA, Hayball MP, Dixon AK. Functional images of hepatic perfusion obtained with dynamic CT. Radiology 1993;188:405–11.[Abstract/Free Full Text]
- Leen E, Angerson WG, Cooke TG, McArdle CS. Prognostic power of Doppler perfusion index in colorectal cancer. Correlation with survival. Ann Surg 1996;223:199–203.[CrossRef][Medline]
- Cuenod CA, Leconte I, Siauve N, Resten A, Dromain C, Poulet B, et al. Early changes in liver perfusion caused by occult metastases in rats: detection with quantitative CT. Radiology 2001;218:556–61.[Abstract/Free Full Text]
- Nott DM, Grime SJ, Yates J, Day DW, Baxter JN, Jenkins SA, et al. Changes in the hepatic perfusion index during the development of experimental hepatic tumours. Br J Surg 1989;76:259–63.[Medline]
- Lee VS, Lavelle MT, Rofsky NM, Laub G, Thomasson DM, Krinsky GA, et al. Hepatic MR imaging with dynamic contrast enhanced isotropic volumetric breath hold examination: feasibility, reproducibility and technical quality. Radiology 2000;215:365–72.[Abstract/Free Full Text]
- Henderson E, Rutt BK, Lee TY. Temporal sampling requirements for the tracer kinetics modelling of breast disease. Magn Reson Imaging 1998;16:1057–73.[CrossRef][Medline]
- Sadek AG, Mohamed FB, Outwater EK, El-Essawy SS, Mitchel DG. Respiratory and postprandial changes in portal flow rate: assessment of phase contrast MR imaging. J Magn Reson Imaging 1996;1:90–3.
- Shuman WP. Liver metastases from colorectal carcinoma: detection with Doppler US-guided measurements of liver blood flow – past, present, future. Radiology 1995;195:9–10.[Free Full Text]
- Fowler RC, Harris KM, Swift SE, Ward M, Greenwood DC. Hepatic Doppler perfusion index: measurement in nine healthy volunteers. Radiology 1998;209:867–71.[Abstract/Free Full Text]
- Materne R, Smith AM, Peeters F, Dehoux JP, Keyeux A, Horsmans Y, et al. Assessment of hepatic perfusion parameters with dynamic MRI. Magn Reson Med 2002;47:135–42.[CrossRef][Medline]
This article has been cited by other articles:
|
|
|
|
 
J H ZHOU, A H LI, L H CAO, H H JIANG, L Z LIU, X Q PEI, and F HAN
Haemodynamic parameters of the hepatic artery and vein can detect liver metastases: assessment using contrast-enhanced ultrasound
Br. J. Radiol.,
February 1, 2008;
81(962):
113 - 119.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
M J White, R L O'Gorman, E M Charles-Edwards, P A Kane, J B Karani, M O Leach, and J J Totman
Parametric mapping of the hepatic perfusion index with gadolinium-enhanced volumetric MRI
Br. J. Radiol.,
February 1, 2007;
80(950):
113 - 120.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A Parkin, K E Goldstone, and P J Robinson
Hepatic perfusion index measured using MRI and scintigraphy
Br. J. Radiol.,
December 1, 2005;
78(936):
1118 - 1118.
[Full Text]
[PDF]
|
|
|
|
|
|
|
C Harvey and M Blomley
Imaging investigation of liver haemodynamics in patients at risk for hepatic metastatic disease
Br. J. Radiol.,
February 1, 2005;
78(926):
103 - 104.
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
