BJR
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
British Journal of Radiology (2006) 79, 578-583
© 2006 British Institute of Radiology
doi: 10.1259/bjr/18842556
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Goh, V
Right arrow Articles by Bartram, C I
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Goh, V
Right arrow Articles by Bartram, C I

Full paper

Quantitative colorectal cancer perfusion measurement by multidetector-row CT: does greater tumour coverage improve measurement reproducibility?

V Goh, MA, MRCP, FRCR1,2, S Halligan, MD, FRCP, FRCR1, L Gartner, MRCP1, P Bassett, MSc1 and C I Bartram, FRCP, FRCS, FRCR1

1 Intestinal Imaging Centre, St Mark's Hospital, Harrow, 2 Paul Strickland Scanner Centre, Mount Vernon Hospital, Northwood, UK

Correspondence: Professor Steve Halligan, Intestinal Imaging Centre, St Mark's Hospital, Watford Road, Harrow, Middlesex HA1 3UJ, UK. Current address: Specialist Radiology, Level 2 Podium, University College Hospital, 235 Euston Road, London NW1 2BU, UK.


   Abstract
Top
 Abstract
Introduction
 Materials and methods
 Results
 Discussion
 References
 
The purpose of this study was to determine if greater z-axis tumour coverage improves the reproducibility of quantitative colorectal cancer perfusion measurements using CT. A 65 s perfusion study was acquired following intravenous contrast administration in 10 patients with proven colorectal cancer using a four-detector row scanner. This was repeated within 48 h using identical technical parameters to allow reproducibility assessment. Quantitative tumour blood volume, blood flow, mean transit time and permeability measurements were determined using commercially available software (Perfusion 3.0; GE Healthcare, Waukesha, WI) for data obtained from a 5 mm z-axis tumour coverage, and from a 20 mm z-axis tumour coverage. Measurement reproducibility was assessed using Bland-Altman statistics, for a 5 mm z-axis tumour coverage, and 20 mm z-axis tumour coverage, respectively. The mean difference (95% limits of agreement) for blood volume, blood flow, mean transit time and permeability were 0.04 (–2.50 to +2.43) ml/100 g tissue; +8.80 (–50.5 to +68.0) ml/100 g tissue/min; –0.99 (–8.19 to +6.20) seconds; and +1.20 (–5.42 to +7.83) ml/100 g tissue/min, respectively, for a 5 mm coverage, and –0.04 (–2.61 to +2.53) ml/100 g tissue; +7.40 (–50.3 to +65.0) ml/100 g tissue/min; –2.46 (–12.61 to +7.69) seconds; and –0.23 (–8.31 to +7.85) ml/100 g tissue/min, respectively, for a 20 mm coverage, indicating similar levels of agreement. In conclusion, increasing z-axis coverage does not improve reproducibility of quantitative colorectal cancer perfusion measurements.


   Introduction
Top
 Abstract
 Introduction
Materials and methods
 Results
 Discussion
 References
 
Perfusion imaging techniques have been advocated increasingly for assessment of tumour response to therapy because antiangiogenic and vascular targeting drugs may not cause tumour shrinkage, despite being therapeutically effective. As a result, conventional response assessment based on changes in tumour size, such as RECIST or WHO criteria [1, 2], may underestimate the effect of these agents [3]. For example, Hurwitz et al reported a 5 month improvement in overall survival in patients with advanced colorectal cancer who had been treated with bevacizumab (Avastin; Genentech, CA), an antivascular endothelial growth factor agent: this was accompanied by a response rate of only 10% using conventional size-based criteria [4].

Tumours are spatially and temporally heterogeneous; at any time point they may recruit 20–85% of the vasculature available to them [5]. Thus functional assessment derived from a single tumour level (or time) may not reflect perfusion in the tumour as a whole. Greater anatomical coverage is now possible with multidetector row CT scanners and this is likely to increase in the future as ever more detector rows become possible. To date, there has been no attempt to determine if increased tumour coverage results in decreased measurement variability for colorectal cancer. We obtained colorectal cancer perfusion measurements from a single 5 mm axial slice, and from 4 contiguous 5 mm slices, producing a z-axis coverage of 20 mm, to determine if reproducibility is enhanced when a greater volume of data is analysed.


   Materials and methods
Top
 Abstract
 Introduction
 Materials and methods
Results
 Discussion
 References
 
Subjects
This prospective study received local ethics committee approval, and written informed consent was obtained from all patients. 10 patients (mean age 67 years, range 39–84 years; 6 male, 4 female) with histologically proven colorectal adenocarcinoma, and attending for a CT staging examination, were recruited prospectively. Of the 10 tumours examined, 5 were located in the rectum and 5 in the sigmoid colon. Of these, 2 were T2, 5 were T3, and 3 were T4 tumours (mean size 6.2 cm, range 5–11.5 cm).

CT scanning
Following a 4 h fast, 1000 ml of water-soluble contrast, 2–4% meglumine and sodium diatrizoate (Gastrografin; Bracco, Milan, Italy), was ingested orally 30 min prior to CT scanning in order to opacify the small bowel, as per normal practice in our institution. With the patient lying supine on the scanner table, an 18 G venous cannula was sited in the antecubital fossa. 20 mg of the spasmolytic hyoscine N butylbromide (Buscopan; Boehringer Ingelheim, Ingelheim am Rheim, Germany) was administered intravenously to reduce bowel peristalsis. Patient movement was minimized by placing a restraining band around the abdomen.

All patients were scanned using a four-detector row CT scanner (Lightspeed Plus; GE Healthcare Technologies, Waukesha, WI). A non-contrast abdominal-pelvic study was performed initially in order to confirm the location of the known colorectal tumour, using the following parameters: slice thickness/interval 10 mm/5 mm, mode HS/speed 30 mm s–1 (pitch 1.5), 120 kV, 180 mA, 0.6 s rotation speed, scan field of view (SFOV) 50 cm, matrix 512 mmx512 mm. The images were then inspected on the CT console by the supervising radiologist, the mid-tumour level identified, the scan location noted and thus used to plan the subsequent dynamic study.

The dynamic studies were performed using contrast and image acquisition protocols in accordance with manufacturer's recommendations. A pump injector (Percupump Touchscreen; EZ-EM, Westbury, NY) was used to inject 100 ml of iopamidol 340 (Niopam 340; Bracco) intravenously at a rate of 5 ml s–1. Four contiguous slices collimated to 5 mm each were obtained at 1 s intervals through the mid-point of the tumour using a "cine mode" (120 kV, 60 mA, SFOV 50 cm, matrix 512 mmx512 mm). Data acquisition commenced 5 s following the start of intravenous injection, to obtain baseline non-contrasted images, and lasted for a total duration of 65 s.

All patients returned within 48 h of the initial study for a second dynamic study specifically to assess measurement reproducibility. Scans were acquired in an identical fashion to the initial study. In particular, the second non-contrast planning scan was compared with that used for the initial study so that the tumour level examined for each study could be matched. Intravenous spasmolytic and contrast were administered exactly as previously and data was acquired using technical parameters identical to the initial dynamic study.

Image analysis
Image analysis was performed by a single radiologist experienced in CT perfusion analysis. All 20 dynamic studies (10 patients; 2 studies each) were analysed on a stand-alone workstation (Advantage 4.1; GE Healthcare Technologies) using commercial software based on deconvolution analysis (Perfusion 3.0; GE Healthcare Technologies). The initial 65 s dynamic study for each patient was loaded into the software (Body tumour, Perfusion 3.0) and a single 5 mm axial slice that best visualized the tumour was chosen from the four axial slices available. A processing threshold of 0–120 Hounsfield Units was selected so that the subsequent analysis appropriately included soft tissue, both unenhanced and enhanced.

The arterial input was determined by manually selecting a circular region of interest (ROI) from the control panel and placing this, using a mouse, within either the iliac or femoral arteries, whichever was best visualized in the imaging plane. Arterial attenuation change was determined over the 65 s acquisition by the software. A time-attenuation curve was generated automatically, and from this the timing of the end of the first pass of contrast could be estimated by visual inspection of the shape of the time-attenuation curve. Subsequent selection of this time point separating the two phases of contrast enhancement, as necessitated by the software program, then permitted generation of the four perfusion parametric maps (blood volume, blood flow, mean transit time and permeability) for all of the tissues within the imaging plane, within the processing threshold selected. This arterial ROI was saved using the software so that the exact same sized ROI could be automatically placed in the same location in subsequent analysis in an attempt to minimize measurement variability due to this confounder.

A ROI was then drawn freehand around the peripheral margin of the tumour using an electronic cursor and mouse. Care was taken to exclude perirectal or pericolonic fat and also intraluminal gas, a process that was facilitated by viewing a cine-loop of the perfusion acquisition in order to gauge the degree and margins of patient movement during acquisition. A time-attenuation curve for the selected tumour tissue and the four perfusion parameters within the tumour ROI were then generated. Mean values for the four tumour perfusion parameters (blood volume, blood flow, mean transit time and permeability) from this single 5 mm axial slice were recorded for each individual patient.

Image analysis was repeated for the remaining three contiguous axial slices in the same manner, recalling the arterial ROI to minimize variability due to this. The tumour ROI was drawn freehand due to morphological differences in each slice. Mean values for all four perfusion parameters were recorded for each of the three contiguous slices. Then, in order to obtain overall mean values for all four perfusion parameters for an equivalent z-axis tumour coverage of 20 mm, values were averaged from all four 5 mm axial slices, and recorded for each individual patient.

Image analysis was performed by the same investigator for the second set of dynamic scans obtained from each of the 10 patients. Analysis was performed exactly as previously, for a single 5 mm axial slice, most similar to the initial analysis, and for all four slices. Mean values for the four tumour perfusion parameters (blood volume, blood flow, mean transit time and permeability) for a z-axis tumour coverage of 5 mm, and for a z-axis tumour coverage of 20 mm were recorded for each individual patient.

Statistical analysis
The mean (standard deviation (SD)) of tumour blood volume, blood flow, mean transit time and permeability measurements from a 5 mm and 20 mm z-axis tumour coverage were determined. Measurement reproducibility was assessed using the Bland Altman test statistic [6, 7]: the mean difference, standard deviation (SD) of the differences, and 95% limits of agreement were calculated for each of the four perfusion parameters (blood volume, blood flow, mean transit time and permeability) for a 5 mm and 20 mm z-axis tumour coverage.


   Results
Top
 Abstract
 Introduction
 Materials and methods
 Results
Discussion
 References
 
The mean value (standard deviation, SD) for each perfusion parameter and the mean difference, SD of the differences, and 95% limits of agreement are summarized in Table 1Go, with corresponding scatter and Bland-Altman agreement plots for blood volume and flow shown in GoFigures 1 and 2Go. Similar limits of agreement were obtained between measurements from a 5 mm and 20 mm z-axis coverage, indicating no improvement in measurement reproducibility.


View this table:
[in this window]
[in a new window]
  		Table 1. 95% limits of agreement analysis showing the mean difference, standard deviation (SD) of differences and 95% limits of agreement for repeated measurements of tumour blood volume, blood flow, mean transit time and permeability for a 5 mm and 20 mm z-axis tumour coverage. The overall mean value (SD) for each perfusion parameter is also shown

 

Figure 1
View larger version (8K):
[in this window]
[in a new window]
  		Figure 1. Blood volume: (a) scatter plot showing the distribution of values with a line of perfect agreement shown; and (b,c) Bland-Altman agreement plots with the mean difference and 95% limits of agreement for a 5 mm ({Delta}) and 20 mm (O) z-axis tumour coverage.

 

Figure 2
View larger version (9K):
[in this window]
[in a new window]
  		Figure 2. Blood flow: (a) scatter plot showing the distribution of values with a line of perfect agreement shown; and (b,c) Bland-Altman agreement plots with the mean difference and 95% limits of agreement for a 5 mm ({Delta}) and 20 mm (O) z-axis tumour coverage.

 

   Discussion
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
References
 
CT perfusion has been advocated to monitor the effects of antivascular and antiangiogenic drugs because this quantitative technique allows in vivo assessment of tumour vascularity [8], and is more robust than techniques such as dynamic contrast enhanced MRI [9]. In vivo functional assessment is advantageous because histological methods that measure angiogenesis, such as microvessel density counting, may not accurately reflect functional tumour vascularity [5]. However, whether assessment of tumour perfusion at a single tumour level, rather than of the whole tumour, is representative and adequate is questionable, particularly in the context of therapeutic assessment.

Multidetector row CT permits a number of contiguous axial images to be acquired at a given tumour level, and z-axis coverage is contingent on the number of detector rows. For example, the maximum tumour coverage achievable for a perfusion study on the four-detector row scanner is 20 mm in the z-axis dimension, consisting of 4 contiguous 5 mm axial slices in cine mode. However, the rapid pace of technological advancement has meant that a 40 mm acquisition in the z-axis dimension, consisting of 8 contiguous 5 mm axial slices, will soon be possible for a perfusion study with the latest 64 detector row scanners. It seems logical to assume that by increasing tumour coverage, a more representative assessment of global tumour perfusion would be obtained, and that this would be less prone to measurement error. However, this has not been proven to date with colorectal cancer.

Reproducibility assessment encompasses quantification of all the intrinsic and extrinsic factors that contribute to measurement variability. These include tumour heterogeneity, CT technique, software variability, and observer variability [10]. While it is not possible to separate these factors, it is possible to minimize effects of certain factors to permit evaluation the effect of the volume of data surveyed on reproducibility. For example, it is possible to minimize the effects of observer variability by using a single experienced observer to obtain all measurements [11], and to minimize effects from the technique and software evaluation by keeping technical factors constant between studies. Thus, we compared the reproducibility of measurements from a 5 mm z-axis tumour coverage with that from 4 contiguous 5 mm slices, the results from which had been averaged to provide information equivalent to from a 20 mm z-axis tumour coverage to identify if any improvement in reproducibility was achieved.

Overall, reproducibility was acceptable for all four perfusion measurements obtained from both a 5 mm and 20 mm tumour coverage, and comparable with previously cited reproducibility levels using CT within the intracranial and extracranial circulation in animal and human studies [12–15]. For example, a variability of 13% and 7%, assessed using analysis of variance, has been quoted for repeated cerebral blood flow and blood volume measurements, respectively, in rabbits [13], while a variability of 14%, 20% and 18% has been quoted for repeated blood flow, blood volume and permeability measurements in the rabbit VX2 tumour [15]. Analysing our data in the same manner using analysis of variance would have resulted in a variability of 23%, 14% and 17% for colorectal blood flow, blood volume and permeability, respectively. Reproducibilty of permeability measurements is also comparable with that reported for dynamic contrast enhanced MRI for a variety of extracranial tumours, where a mean difference of 0.03, 95% confidence interval of –0.04 to +0.06, and coefficient of variation of 29% was reported for log transformed values of ktrans [16].

With reference to published data, the differences between repeated blood flow measurements that we observed may be sufficiently small to be overwhelmed by the change in perfusion induced therapeutically by antivascular targeting and antiangiogenic agents being evaluated currently. For example, dynamic contrast enhanced (DCE)-MRI data from a Phase I study of combretastatin, a vascular targeting agent, have shown a significant group mean measurement change of 37% after drug administration [17]. Likewise, perfusion CT data from a Phase I study of bevacizumab have shown mean blood flow changes of the order of 40% [18], while data from a study of the effects of chemoradiation have shown a mean blood flow changes of 62% [19]. However, for treatments that produce a smaller effect on tumour vascularity, some caution must be applied to the in the interpretation of results, as measurement change may remain within the range of measurement variability. On an individual patient basis, there may be a role for assessing individual intrapatient measurement variability, so that any response may be interpreted in the light of this. An alternative would be to use surrogates that show better reproducibility, for example semi-quantitative measurements such as peak enhancement, or standardized perfusion value. Further evaluation of these strategies is required.

Reproducibility did not improve with increasing tumour z-axis coverage. There may be several explanations why data acquisition from a larger tumour volume did not improve reproducibility. Most obviously, it may be the case that colorectal spatial heterogeneity is insufficient to influence data acquisition once the tumour volume has achieved 5 mm in its z-axis dimension. This is in contrast to preliminary data from lung tumours that suggests improved reproducibility with increasing z-axis acquisition [20]. However, the lung tumours studied were large heterogeneous tumours with a necrotic centre, unlike the colorectal tumours within this study, and thus spatial heterogeneity may have been a major factor contributing to measurement variability in these tumours.

It is also possible that the variation in tumour morphology over time encountered with colorectal cancer may be partly responsible for the lack of improvement observed. For example, bowel tumours might change shape and position as a result of peristalsis. Assessment of reproducibility from a single 5 mm-tumour level is straightforward, as it is usually possible to ensure comparable tumour levels on both sets of scans when choosing from the 4 slices available. However, we found that ensuring the same tumour level was assessed on all 4 slices was more difficult, because there was no leeway in post-acquisition slice selection. Tumour shape and position were unlikely to be identical on both sets of scans, despite care in scan acquisition, including administration of an antiperistaltic agent.

It is possible that the quantitative software analysis method we used may have contributed to the lack of improvement found with increased tumour coverage. While the ROI used to define the arterial input may be saved using the software for future use, to minimize variability from ROI placement in subsequent analyses, due to variation in tumour shape between studies, a ROI has to be drawn freehand around the tumour on the parametric perfusion maps, which are generated automatically by the software to obtain the quantitative parameters. Each colour pixel on these maps represents an individual quantitative value of the parameter measured (blood flow, blood volume, mean transit time or permeability) and a mean parameter value is generated for the ROI drawn. It is inevitable that some variability will be introduced when drawing a ROI (though this can be reduced by using the same unblinded observer to create each ROI). Finally, our patient numbers were small and inevitably reduced the power of our experiment, but this was a pragmatic consideration based on a natural reluctance to unnecessarily expose patients to ionizing radiation on two separate occasions.

In summary, we found that increasing z-axis tumour coverage did not improve reproducibility of measurements of colorectal tumour perfusion. Measurement reproducibility remains clinically acceptable with single level measurements.

This research was supported by a grant from the Royal College of Radiologists, London, UK.

Received for publication August 24, 2005. Revision received November 7, 2005. Accepted for publication November 24, 2005.


   References
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 

  1. Therasse P, Arbuck SG, Eisenhauer EA, Wanders J, Kaplan RS, Rubenstein L, et al. New guidelines to evaluate the response to treatment in solid tumors. J Natl Cancer Inst 2000;92:205–16.[Abstract/Free Full Text]
  2. World Health Organisation. WHO hand book for reporting results of cancer treatment. WHO offset publication 1979;no 48. Geneva, Switzerland
  3. Tozer GM. Measuring tumour vascular response to antivascular and antiangiogenic drugs. Br J Radiol 2003;76:S23–S35.[Abstract/Free Full Text]
  4. Hurwitz H, Fehrenbacher L, Novotny W, Cartwright T, Hainsworth J, Heim W, et al. Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer. N Engl J Med 2004;350:2335–42.[Abstract/Free Full Text]
  5. Endrich B, Vaupel P. The role of the microcirculation in the treatment of malignant tumors: facts and fiction. In: Molls M, Vaupel P, editors. Blood perfusion and microenvironment of human tumors: implications for clinical radiooncology. Berlin, Germany: Springer-Verlag, 2001;19–39
  6. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–10.[CrossRef][Medline]
  7. Bland JM, Altman DG. Applying the right statistics: analyses of measurement studies. Ultrasound Obstet Gynecol 2003;22:85–93.[Medline]
  8. Miles KA, Griffiths MR. Perfusion CT: a worthwhile enhancement? Br J Radiol 2003;76:220–31.[Free Full Text]
  9. Goh V, Halligan S, Balmer JA, Bartram CI. Colorectal cancer perfusion measurement with multi-detector computed tomography: measurement reproducibility and clinical implications. In: Radiological Society of North America 90th Scientific Assembly and Annual Meeting Program; 2004. Chicago, USA: RSNA, 2004
  10. Bland M. An introduction to medical statistics, 3rd edn. Oxford, UK: Oxford University Press, 2003
  11. Goh V, Halligan S, Hugill JA, Bassett P, Bartram CI. Quantitative assessment of colorectal cancer perfusion using MDCT: inter and intra-observer agreement. AJR Am J Roentgenol 2005;185:225–31.[Abstract/Free Full Text]
  12. Nabavi DG, Cenic A, Dool J, Smith RM, Espinosa F, Craen R, et al. Quantitative assessment of cerebral hemodynamics using CT: stability, accuracy, and precision studies in dogs. J Comput Assist Tomogr 1999;23:506–15.[CrossRef][Medline]
  13. Cenic A, Nabavi DG, Craen RA, Gelb AW, Lee TY. A CT method to measure hemodynamics in brain tumors: validation and application of cerebral flow maps. AJNR Am J Neuroradiol 2000;21:462–70.[Abstract/Free Full Text]
  14. Gillard JH, Antoun NM, Burnet NG, Pickard JD. Reproducibility of quantitative CT perfusion imaging. Br J Radiol 2001;74:552–5.[Abstract/Free Full Text]
  15. Purdie TG, Henderson E, Lee TY. Functional CT imaging of angiogenesis in rabbit VX2 soft tissue tumour. Phys Med Biol 2001;46:3161–75.[CrossRef][Medline]
  16. Galbraith SM, Lodge MA, Taylor NJ, Rustin GJ, Bentzen S, Stirling JJ, et al. Reproducibility of dynamic contrast enhanced MRI in human muscle and tumours: comparison of quantitative and semi-quantitative analysis. NMR Biomed 2002;15:132–42.[CrossRef][Medline]
  17. Galbraith SM, Maxwell RJ, Lodge MA, Tozer GM, Wilson J, Taylor NJ, et al. Combretastatin A4 Phosphate has tumor antivascular activity in rat and man as demonstrated by dynamic magnetic resonance imaging. J Clin Oncol 2003;21:2831–42.[Abstract/Free Full Text]
  18. Willett CG, Boucher Y, di Tomaso E, Duda DG, Munn LL, Tong RT, et al. Direct evidence that the VEGF-specific antibody bevacizumab has antivascular effects in human rectal cancer. Nature Med 2004;10:145–7.[CrossRef][Medline]
  19. Sahani DV, Kalva SP, Hamberg LM, Hahn PF, Willett CG, Saini S, et al. Assessing tumor perfusion and treatment response in rectal cancer with multisection CT: initial observations. Radiology 2005;234:785–92.[Abstract/Free Full Text]
  20. Ng QS, Goh V, Klotz E, Fichte H, Carnell D, Saunders M, et al. Single versus four-level tumour perfusion measurement using helical dynamic contrast enhanced MDCT: effect on measurement agreement and potential for whole tumour assessment. In: Proceedings for European Congress of Radiology 2005; Vienna, Austria. Eur Radiol 2005;15(Suppl. 1):414




This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Goh, V
Right arrow Articles by Bartram, C I
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Goh, V
Right arrow Articles by Bartram, C I

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
BJR DMFR IMAGING  ALL BIR JOURNALS