Journal of Medical Imaging and Radiation Oncology 58 (2014) 32–37 bs_bs_banner
RADIOLO GY —O R I G I N AL A R T I C L E
Solid focal liver lesion characterisation with apparent diffusion coefficient ratios Tom Sutherland, Emma Steele, Frans van Tonder and Kelvin Yap Medical Imaging Department, St Vincents Hospital, Melbourne, Victoria, Australia
T Sutherland MBBS MMed FRANZCR; E Steele Ba App Sci (Medical Radiations) Grad Dip(Magnetic Resonance); F van Tonder MBChB; K Yap MBBS MMed FRACP FRANZCR. Correspondence Dr Tom Sutherland, Medical Imaging Department, St Vincents Hospital, 41 Victoria Parade, Fitzroy 3065, Australia. Email: [email protected] All authors agree with the text. Conflict of interest: No conflict of interest. Submitted 5 March 2013; accepted 25 May 2013. doi:10.1111/1754-9485.12087
Abstract Introduction: Non-invasive characterisation of focal liver lesions using diffusion-weighted imaging (DWI) has been heavily investigated and has shown substantial overlap between benign and malignant lesions. We have calculated a ratio of lesion to normal liver to determine if it improves accuracy for correct categorisation. Method: All hepatic MRI studies performed between 1st April 2009 and 26th September 2011 were retrospectively reviewed. Patients with solid focal liver lesions in whom a diagnosis could be established and had lesions over 10 mm were included. Haemangiomas, cysts and patients with chronic liver disease were excluded. Apparent diffusion coefficient (ADC) values were calculated for each lesion and adjacent normal liver on breath hold DWI. Results: Two hundred fifty-eight studies were performed and 206 were excluded leaving 52 scans and 58 lesions of which 47 were benign and 11 were malignant. The mean ADC value for benign lesions was 1196.6 (two standard deviations (2SD) = ⫾399.9) and of benign liver 1101.5 (2SD = ⫾329.8) with a ratio of benign lesion to benign liver of 1.1005 (2SD = ⫾0.3783). The mean ADC of malignant lesions was 1153.0 (2SD = ⫾604.9) and malignant liver of 1080.7 (2SD = ⫾533.4) giving a malignant lesion to malignant liver ratio of 1.0890 (2SD = ⫾0.4975). None of these results were statistically significant (all P > 0.5). Conclusion: DWI is unable to reliably differentiate solid benign lesions from solid malignant lesions. Key words: diffusion-weighted imaging; lesion characterisation; liver imaging.
Introduction
Methods
The characterisation of focal liver lesions is a daily challenge faced by radiologists. The accurate non-invasive categorisation of focal liver lesions as either benign or malignant is vital for patient management. Diffusionweighted imaging (DWI) has been heavily investigated for lesion characterisation, and results have been disappointing with substantial overlap in apparent diffusion coefficient (ADC) value between solid benign and malignant lesions. Our study examines the ADC of focal liver lesions and normal adjacent liver to determine if this ratio can more accurately differentiate benign from malignant pathologies.
All hepatic MRI examinations performed in a tertiary hospital between 1st April 2009 and 26th September 2011 were retrospectively reviewed. Patients with solid lesion(s) 10 mm or greater where the diagnosis was confirmed via serial imaging or pathology were included. Patients without lesions, or lesions that were conclusively cysts or haemangiomas were excluded. Patients with lesions who had been treated with chemotherapy were excluded. Patients with chronic liver disease were excluded. Patients with lesions that were of uncertain aetiology, due to inadequate gold standard were also excluded. Patients with multiple examinations over the study period only had a single study included for analysis.
32
© 2013 The Royal Australian and New Zealand College of Radiologists
DWI for liver lesion characterisation
All included studies were reviewed by a single radiologist (TS) with 4 years of experience reporting hepatic MRI, who identified all lesions over 10 mm by segment, series and image number, and correlated with histology and subsequent follow-up examinations. All lesions fulfilling inclusion criteria were then reviewed on a commercially available workstation (Siemens Leonardo Workstation) by a medical imaging technologist (ES), and the ADC value of each lesion was recorded, along with the ADC value of adjacent normal liver taking care to avoid major vessels, hepatic periphery, artefacts and areas of necrosis. Accounting for these factors, the largest region of interest (ROI) was used for each lesion, and an identical size ROI was used for normal liver. A ratio of ‘lesion ADC’ to normal ‘liver ADC’ was then calculated. Benign results were compared with malignant results using a two-sided Student’s t-test.
DWI technique All scans were performed on a 1.5 T Siemens Avanto (Siemens Medical Solution, Erlangen, Germany) using breath hold and the following factors: TE 73 msec, TR 2300 msec, eight averages, three directions, 23 slices 8-mm thick with FOV 370 mm using a 16 channel body array coil and b values of 100, 400 and 800.
Gold standard Five benign lesions had histologic proof comprising one hyalinised scar, one hepatocellular adenoma (HCA),
three focal nodular hyperplasia (FNH) and one focal steatosis. Of the three FNH with histology, two were biopsied because they were atypical (one had no arterial enhancement on MRI and another grew from 2.9– 4.7 cm). The third was resected as it was symptomatic at 10.2 cm and was causing capsular stretch and compressing vascular structures. The remaining FNHs were diagnosed based upon established typical MRI features1 and in all but two cases this included using Gadoxetic disodium (Primovist [Eovist in the USA]; Bayer Schering Pharma, Berlin, Germany), which has been demonstrated to have a high sensitivity and specificity for diagnosing FNH.2 The two FNHs that were not imaged using Gadoxetic acid were both stable at follow up of over 24 months. Malignant lesions were confirmed histologically in all patients. All malignant lesions were solitary except for one patient with two lesions. Only one of these was histologically sampled as they both had identical imaging characteristics.
Results Between 1st April 2009 and 30th September 2011, a total of 258 hepatic MRI studies were performed. Two hundred scans were excluded: 107 (53.5%) due to chronic liver disease; 22 (11.0%) as no focal lesion was present; 10 (5.0%) with no lesion over 10 mm; 24 (12.0%) as only haemangiomas were present; and 37 (18.0%) for other reasons including vascular anomalies other than haemangioma, Caroli disease, Budd Chiari, peliosis and incomplete data set due to abandoned scan.
Fig. 1. Box and whisker plots of results divided in quartiles, solid line representing the mean. (a) ADC value of benign and malignant lesions. (b) ADC value of normal liver adjacent to lesions. (c) Ratio of lesion to liver for benign and malignant pathologies.
© 2013 The Royal Australian and New Zealand College of Radiologists
33
T Sutherland et al.
a
c
b
d
Fig. 2. Hepatic cholangiocarcinoma is outlined by arrows in Figure 2a on a b50 DWI image. Figure 2b is at b 400, and Figure 2c is b 800, whereas Figure 2d is an ADC map with a ROI over the lesion and a second ROI over normal liver in segment 2/3.
Fifty-eight scans met inclusion criteria, with 52 scans remaining once follow-up studies were excluded. Fortythree (83%) patients were females and nine were males with a mean age of 45.6 years. A total of 58 lesions were included of which 47 (81%) were benign and 11 (19%) were malignant. Of the benign lesions, 40 were FNH, four hepatocellular adenoma, two focal steatosis and one case of hyalinised scar. The malignant lesions were nine metastases (six colorectal primaries, one carcinoid primary, one fibrolamellar HCC metastasis and one ampullary adenocarcinoma metastasis) and two were mass forming intrahepatic cholangiocarcinomas. Mean lesion size was 26.2 mm (range 11–102 mm). Mean benign lesion size was 26.5 mm (range 11– 102 mm) and mean malignant lesion size was 24.7 mm (range 11–40 mm). The mean ADC value for benign lesions was 1196.6 (two standard deviations (2SD) = ⫾399.9) and of benign liver 1101.5 (2SD = ⫾329.8) with a ratio of benign lesion 34
to benign liver of 1.1005 (2SD = ⫾0.3783). The mean ADC of malignant lesions was 1153.0 (2SD = ⫾604.9) and malignant liver of 1080.7 (2SD = ⫾533.4) giving a malignant lesion to malignant liver ratio of 1.0890 (2SD = ⫾0.4975). None of these results were statistically significant (all P > 0.5). The results have been presented graphically in Figure 1. Example lesions are presented as Figure 2 and Figure 3.
Discussion Technological advances such as parallel imaging have allowed DWI to be applied to hepatic imaging ,and it has been investigated for both lesion detection3,4 and lesion characterisation.3,5–12 Although the usefulness of DWI for lesion detection has been established, the role for lesion characterisation is still uncertain. Although many studies have demonstrated that DWI can differentiate benign from malignant lesions, most of these studies have © 2013 The Royal Australian and New Zealand College of Radiologists
DWI for liver lesion characterisation
a
b
c
d
Fig. 3. A focal nodular hyperplasia in the right lobe of liver (Fig. 3a) on b 50 is outlined by arrows. It is seen on b400 in Figure 3b and b 800 in Figure 3c, whereas the ADC map in Figure 3d contains a ROI over the lesion avoiding the scar and a ROI over adjacent normal liver.
included cysts and haemangiomas which, due to their high ADC values, substantially skews the results. Solid benign lesions have typically been underrepresented in previous studies, and we have found no other study comparing a large number of solid benign lesions with solid malignant lesions using identical image acquisition technique. Given that cysts and haemangiomas can usually be confidently diagnosed on hepatic MRI, we chose to excluded them and focus on solid benign lesions that are more of a diagnostic challenge. In our study, the cellularity of FNH resulted in low ADC values and substantial overlap with malignant lesion ADC. The low ADC of FNH and HCA has been demonstrated by other investigators13 who did not directly compare results with a malignant group. Most other investigators have attempted to determine an ADC threshold that can be used to differentiate benign and malignant lesions. However, the ADC value of normal liver is variable6,9,14 and is effected by the b values used, the acquisition technique (respiratory triggered, vs. free breathing vs. breath hold),15,16 shot number,17 pulse triggering,5 the anatomical location,5,18 portal venous blood flow related to post-prandial state,19 artefacts and © 2013 The Royal Australian and New Zealand College of Radiologists
hepatic fibrosis.20 These factors make the application of a single ADC threshold for characterisation unreliable. The choice of b values varies between studies, with low b values being affected by microcapillary perfusion and high b values having reduced signal to noise ratio (SNR). The use of more b values may increase the accuracy of calculated ADC values but is more time consuming, which limits its daily application in busy practices. The use of respiratory triggering can increase the SNR at high b values by allowing more acquisitions to be obtained6 and allows smaller lesions to be examined21; however, the acquisition time is substantially longer.15 Despite the limitations of breath hold DWI, it is considered acceptable to use in busy practices.22 We used the ADC of normal liver near a focal lesion as an internal control to calculate a ratio and hopefully negate the effect of inherent ADC variability. Other investigators10 have used a ratio to attempt to detect differences between HCC and metastases while accounting for the effects of fibrosis, but we are unaware of any investigators using a ratio as an internal control for differentiating benign and malignant lesions in a noncirrhotic liver. 35
T Sutherland et al.
Our study found that solid benign lesions have low ADC values and substantial overlap with malignant lesions. Furthermore, the use of an internal control of normal liver did not allow accurate differentiation of benign from malignant lesions. Agnello et al.13 recently examined a large number of FNH and HCA and also calculated a ratio compared with normal liver. Their findings correlate well with ours that solid benign lesions are relatively restricted compared with normal liver. Given the similarity of ADC ratio between benign and malignant lesions, and the inherent variability of ADC derived from breath hold acquisitions, respiratory triggered sequences with higher SNR may be required to detect a small difference if it exists. The application of more b values may also help detect subtle differences. A limitation of our study is the relatively small number of malignant lesions. However, the lesions showed ADC overlap with benign pathologies and while a larger series may show occasional outliers, ADC values by themselves will most likely remain poor discriminators of malignancy. Another limitation is the relatively high proportion of female patients in our cohort. This is due to MRI with hepatocellular contrast agents being used to confirm the diagnosis of FNH, which is most common in females, and was the most common solid benign lesion in our group. To our knowledge, there is no study reporting different hepatic ADC values between males and females and so this limitation is unlikely to skew our results.
Conclusion Liver MRI is a well-established tool in the characterisation of liver lesions. The application of DWI in the characterisation of solid lesions has been disappointing in the past. This study further establishes that the careful use of lesion ADC to normal liver ratios cannot distinguish solid malignant lesions from solid benign lesions.
References 1. Ferlicot S, Kobeiter H, Tran Van Nhieu J et al. MRI of atypical focal nodular hyperplasia of the liver: radiology-pathology correlation. AJR Am J Roentgenol 2004; 182: 1227–31. 2. Zech CJ, Grazioli L, Breuer J, Reiser MF, Schoenberg SO. Diagnostic performance and description of morphological features of focal nodular hyperplasia in Gd-EOB-DTPA-enhanced liver magnetic resonance imaging: results of a multicenter trial. Invest Radiol 2008; 43: 504–11. 3. Parikh T, Drew SJ, Lee VS et al. Focal liver lesion detection and characterization with diffusion-weighted MR imaging: comparison with standard breath-hold T2-weighted imaging. Radiology 2008; 246: 812–22. 4. Coenegrachts K, Delanote J, Ter Beek L et al. Improved focal liver lesion detection: comparison of single-shot diffusion-weighted echoplanar and
36
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
single-shot T2 weighted turbo spin echo techniques. Br J Radiol 2007; 80: 524–31. Bruegel M, Holzapfel K, Gaa J et al. Characterization of focal liver lesions by ADC measurements using a respiratory triggered diffusion-weighted single-shot echo-planar MR imaging technique. Eur Radiol 2008; 18: 477–85. Gourtsoyianni S, Papanikolaou N, Yarmenitis S, Maris T, Karantanas A, Gourtsoyiannis N. Respiratory gated diffusion-weighted imaging of the liver: value of apparent diffusion coefficient measurements in the differentiation between most commonly encountered benign and malignant focal liver lesions. Eur Radiol 2008; 18: 486–92. Sandrasegaran K, Akisik FM, Lin C, Tahir B, Rajan J, Aisen AM. The value of diffusion-weighted imaging in characterizing focal liver masses. Acad Radiol 2009; 16: 1208–14. Onur MR, Cicekci M, Kayali A, Poyraz AK, Kocakoc E. The role of ADC measurement in differential diagnosis of focal hepatic lesions. Eur J Radiol 2012; 81: e171–6. Moteki T, Horikoshi H, Oya N, Aoki J, Endo K. Evaluation of hepatic lesions and hepatic parenchyma using diffusion-weighted reordered turboFLASH magnetic resonance images. J Magn Reson Imaging 2002; 15: 564–72. Sun XJ, Quan XY, Huang FH, Xu YK. Quantitative evaluation of diffusion-weighted magnetic resonance imaging of focal hepatic lesions. World J Gastroenterol 2005; 11: 6535–7. Kilickesmez O, Bayramoglu S, Inci E, Cimilli T. Value of apparent diffusion coefficient measurement for discrimination of focal benign and malignant hepatic masses. J Med Imaging Radiat Oncol 2009; 53: 50–5. Demir OI, Obuz F, Sagol O, Dicle O. Contribution of diffusion-weighted MRI to the differential diagnosis of hepatic masses. Diagn Interv Radiol 2007; 13: 81–6. Agnello F, Ronot M, Valla DC, Sinkus R, Van Beers BE, Vilgrain V. High-b-value diffusion-weighted MR imaging of benign hepatocellular lesions: quantitative and qualitative analysis. Radiology 2012; 262: 511–9. Taouli B, Martin AJ, Qayyum A et al. Parallel imaging and diffusion tensor imaging for diffusion-weighted MRI of the liver: preliminary experience in healthy volunteers. AJR Am J Roentgenol 2004; 183: 677–80. Kwee TC, Takahara T, Koh DM, Nievelstein RA, Luijten PR. Comparison and reproducibility of ADC measurements in breathhold, respiratory triggered, and free-breathing diffusion-weighted MR imaging of the liver. J Magn Reson Imaging 2008; 28: 1141–8. Taouli B, Sandberg A, Stemmer A et al. Diffusion-weighted imaging of the liver: comparison of navigator triggered and breathhold acquisitions. J Magn Reson Imaging 2009; 30: 561–8.
© 2013 The Royal Australian and New Zealand College of Radiologists
DWI for liver lesion characterisation
17. Shiehmorteza M, Sirlin CB, Wolfson T et al. Effect of shot number on the calculated apparent diffusion coefficient in phantoms and in human liver in diffusion-weighted echo-planar imaging. J Magn Reson Imaging 2009; 30: 547–53. 18. Nasu K, Kuroki Y, Sekiguchi R, Kazama T, Nakajima H. Measurement of the apparent diffusion coefficient in the liver: is it a reliable index for hepatic disease diagnosis? Radiat Med 2006; 24: 438–44. 19. Hollingsworth KG, Lomas DJ. Influence of perfusion on hepatic MR diffusion measurement. NMR Biomed 2006; 19: 231–5.
© 2013 The Royal Australian and New Zealand College of Radiologists
20. Taouli B, Tolia AJ, Losada M et al. Diffusion-weighted MRI for quantification of liver fibrosis: preliminary experience. AJR Am J Roentgenol 2007; 189: 799–806. 21. Holzapfel K, Bruegel M, Eiber M et al. Characterization of small (
RADIOLO GY —O R I G I N AL A R T I C L E
Solid focal liver lesion characterisation with apparent diffusion coefficient ratios Tom Sutherland, Emma Steele, Frans van Tonder and Kelvin Yap Medical Imaging Department, St Vincents Hospital, Melbourne, Victoria, Australia
T Sutherland MBBS MMed FRANZCR; E Steele Ba App Sci (Medical Radiations) Grad Dip(Magnetic Resonance); F van Tonder MBChB; K Yap MBBS MMed FRACP FRANZCR. Correspondence Dr Tom Sutherland, Medical Imaging Department, St Vincents Hospital, 41 Victoria Parade, Fitzroy 3065, Australia. Email: [email protected] All authors agree with the text. Conflict of interest: No conflict of interest. Submitted 5 March 2013; accepted 25 May 2013. doi:10.1111/1754-9485.12087
Abstract Introduction: Non-invasive characterisation of focal liver lesions using diffusion-weighted imaging (DWI) has been heavily investigated and has shown substantial overlap between benign and malignant lesions. We have calculated a ratio of lesion to normal liver to determine if it improves accuracy for correct categorisation. Method: All hepatic MRI studies performed between 1st April 2009 and 26th September 2011 were retrospectively reviewed. Patients with solid focal liver lesions in whom a diagnosis could be established and had lesions over 10 mm were included. Haemangiomas, cysts and patients with chronic liver disease were excluded. Apparent diffusion coefficient (ADC) values were calculated for each lesion and adjacent normal liver on breath hold DWI. Results: Two hundred fifty-eight studies were performed and 206 were excluded leaving 52 scans and 58 lesions of which 47 were benign and 11 were malignant. The mean ADC value for benign lesions was 1196.6 (two standard deviations (2SD) = ⫾399.9) and of benign liver 1101.5 (2SD = ⫾329.8) with a ratio of benign lesion to benign liver of 1.1005 (2SD = ⫾0.3783). The mean ADC of malignant lesions was 1153.0 (2SD = ⫾604.9) and malignant liver of 1080.7 (2SD = ⫾533.4) giving a malignant lesion to malignant liver ratio of 1.0890 (2SD = ⫾0.4975). None of these results were statistically significant (all P > 0.5). Conclusion: DWI is unable to reliably differentiate solid benign lesions from solid malignant lesions. Key words: diffusion-weighted imaging; lesion characterisation; liver imaging.
Introduction
Methods
The characterisation of focal liver lesions is a daily challenge faced by radiologists. The accurate non-invasive categorisation of focal liver lesions as either benign or malignant is vital for patient management. Diffusionweighted imaging (DWI) has been heavily investigated for lesion characterisation, and results have been disappointing with substantial overlap in apparent diffusion coefficient (ADC) value between solid benign and malignant lesions. Our study examines the ADC of focal liver lesions and normal adjacent liver to determine if this ratio can more accurately differentiate benign from malignant pathologies.
All hepatic MRI examinations performed in a tertiary hospital between 1st April 2009 and 26th September 2011 were retrospectively reviewed. Patients with solid lesion(s) 10 mm or greater where the diagnosis was confirmed via serial imaging or pathology were included. Patients without lesions, or lesions that were conclusively cysts or haemangiomas were excluded. Patients with lesions who had been treated with chemotherapy were excluded. Patients with chronic liver disease were excluded. Patients with lesions that were of uncertain aetiology, due to inadequate gold standard were also excluded. Patients with multiple examinations over the study period only had a single study included for analysis.
32
© 2013 The Royal Australian and New Zealand College of Radiologists
DWI for liver lesion characterisation
All included studies were reviewed by a single radiologist (TS) with 4 years of experience reporting hepatic MRI, who identified all lesions over 10 mm by segment, series and image number, and correlated with histology and subsequent follow-up examinations. All lesions fulfilling inclusion criteria were then reviewed on a commercially available workstation (Siemens Leonardo Workstation) by a medical imaging technologist (ES), and the ADC value of each lesion was recorded, along with the ADC value of adjacent normal liver taking care to avoid major vessels, hepatic periphery, artefacts and areas of necrosis. Accounting for these factors, the largest region of interest (ROI) was used for each lesion, and an identical size ROI was used for normal liver. A ratio of ‘lesion ADC’ to normal ‘liver ADC’ was then calculated. Benign results were compared with malignant results using a two-sided Student’s t-test.
DWI technique All scans were performed on a 1.5 T Siemens Avanto (Siemens Medical Solution, Erlangen, Germany) using breath hold and the following factors: TE 73 msec, TR 2300 msec, eight averages, three directions, 23 slices 8-mm thick with FOV 370 mm using a 16 channel body array coil and b values of 100, 400 and 800.
Gold standard Five benign lesions had histologic proof comprising one hyalinised scar, one hepatocellular adenoma (HCA),
three focal nodular hyperplasia (FNH) and one focal steatosis. Of the three FNH with histology, two were biopsied because they were atypical (one had no arterial enhancement on MRI and another grew from 2.9– 4.7 cm). The third was resected as it was symptomatic at 10.2 cm and was causing capsular stretch and compressing vascular structures. The remaining FNHs were diagnosed based upon established typical MRI features1 and in all but two cases this included using Gadoxetic disodium (Primovist [Eovist in the USA]; Bayer Schering Pharma, Berlin, Germany), which has been demonstrated to have a high sensitivity and specificity for diagnosing FNH.2 The two FNHs that were not imaged using Gadoxetic acid were both stable at follow up of over 24 months. Malignant lesions were confirmed histologically in all patients. All malignant lesions were solitary except for one patient with two lesions. Only one of these was histologically sampled as they both had identical imaging characteristics.
Results Between 1st April 2009 and 30th September 2011, a total of 258 hepatic MRI studies were performed. Two hundred scans were excluded: 107 (53.5%) due to chronic liver disease; 22 (11.0%) as no focal lesion was present; 10 (5.0%) with no lesion over 10 mm; 24 (12.0%) as only haemangiomas were present; and 37 (18.0%) for other reasons including vascular anomalies other than haemangioma, Caroli disease, Budd Chiari, peliosis and incomplete data set due to abandoned scan.
Fig. 1. Box and whisker plots of results divided in quartiles, solid line representing the mean. (a) ADC value of benign and malignant lesions. (b) ADC value of normal liver adjacent to lesions. (c) Ratio of lesion to liver for benign and malignant pathologies.
© 2013 The Royal Australian and New Zealand College of Radiologists
33
T Sutherland et al.
a
c
b
d
Fig. 2. Hepatic cholangiocarcinoma is outlined by arrows in Figure 2a on a b50 DWI image. Figure 2b is at b 400, and Figure 2c is b 800, whereas Figure 2d is an ADC map with a ROI over the lesion and a second ROI over normal liver in segment 2/3.
Fifty-eight scans met inclusion criteria, with 52 scans remaining once follow-up studies were excluded. Fortythree (83%) patients were females and nine were males with a mean age of 45.6 years. A total of 58 lesions were included of which 47 (81%) were benign and 11 (19%) were malignant. Of the benign lesions, 40 were FNH, four hepatocellular adenoma, two focal steatosis and one case of hyalinised scar. The malignant lesions were nine metastases (six colorectal primaries, one carcinoid primary, one fibrolamellar HCC metastasis and one ampullary adenocarcinoma metastasis) and two were mass forming intrahepatic cholangiocarcinomas. Mean lesion size was 26.2 mm (range 11–102 mm). Mean benign lesion size was 26.5 mm (range 11– 102 mm) and mean malignant lesion size was 24.7 mm (range 11–40 mm). The mean ADC value for benign lesions was 1196.6 (two standard deviations (2SD) = ⫾399.9) and of benign liver 1101.5 (2SD = ⫾329.8) with a ratio of benign lesion 34
to benign liver of 1.1005 (2SD = ⫾0.3783). The mean ADC of malignant lesions was 1153.0 (2SD = ⫾604.9) and malignant liver of 1080.7 (2SD = ⫾533.4) giving a malignant lesion to malignant liver ratio of 1.0890 (2SD = ⫾0.4975). None of these results were statistically significant (all P > 0.5). The results have been presented graphically in Figure 1. Example lesions are presented as Figure 2 and Figure 3.
Discussion Technological advances such as parallel imaging have allowed DWI to be applied to hepatic imaging ,and it has been investigated for both lesion detection3,4 and lesion characterisation.3,5–12 Although the usefulness of DWI for lesion detection has been established, the role for lesion characterisation is still uncertain. Although many studies have demonstrated that DWI can differentiate benign from malignant lesions, most of these studies have © 2013 The Royal Australian and New Zealand College of Radiologists
DWI for liver lesion characterisation
a
b
c
d
Fig. 3. A focal nodular hyperplasia in the right lobe of liver (Fig. 3a) on b 50 is outlined by arrows. It is seen on b400 in Figure 3b and b 800 in Figure 3c, whereas the ADC map in Figure 3d contains a ROI over the lesion avoiding the scar and a ROI over adjacent normal liver.
included cysts and haemangiomas which, due to their high ADC values, substantially skews the results. Solid benign lesions have typically been underrepresented in previous studies, and we have found no other study comparing a large number of solid benign lesions with solid malignant lesions using identical image acquisition technique. Given that cysts and haemangiomas can usually be confidently diagnosed on hepatic MRI, we chose to excluded them and focus on solid benign lesions that are more of a diagnostic challenge. In our study, the cellularity of FNH resulted in low ADC values and substantial overlap with malignant lesion ADC. The low ADC of FNH and HCA has been demonstrated by other investigators13 who did not directly compare results with a malignant group. Most other investigators have attempted to determine an ADC threshold that can be used to differentiate benign and malignant lesions. However, the ADC value of normal liver is variable6,9,14 and is effected by the b values used, the acquisition technique (respiratory triggered, vs. free breathing vs. breath hold),15,16 shot number,17 pulse triggering,5 the anatomical location,5,18 portal venous blood flow related to post-prandial state,19 artefacts and © 2013 The Royal Australian and New Zealand College of Radiologists
hepatic fibrosis.20 These factors make the application of a single ADC threshold for characterisation unreliable. The choice of b values varies between studies, with low b values being affected by microcapillary perfusion and high b values having reduced signal to noise ratio (SNR). The use of more b values may increase the accuracy of calculated ADC values but is more time consuming, which limits its daily application in busy practices. The use of respiratory triggering can increase the SNR at high b values by allowing more acquisitions to be obtained6 and allows smaller lesions to be examined21; however, the acquisition time is substantially longer.15 Despite the limitations of breath hold DWI, it is considered acceptable to use in busy practices.22 We used the ADC of normal liver near a focal lesion as an internal control to calculate a ratio and hopefully negate the effect of inherent ADC variability. Other investigators10 have used a ratio to attempt to detect differences between HCC and metastases while accounting for the effects of fibrosis, but we are unaware of any investigators using a ratio as an internal control for differentiating benign and malignant lesions in a noncirrhotic liver. 35
T Sutherland et al.
Our study found that solid benign lesions have low ADC values and substantial overlap with malignant lesions. Furthermore, the use of an internal control of normal liver did not allow accurate differentiation of benign from malignant lesions. Agnello et al.13 recently examined a large number of FNH and HCA and also calculated a ratio compared with normal liver. Their findings correlate well with ours that solid benign lesions are relatively restricted compared with normal liver. Given the similarity of ADC ratio between benign and malignant lesions, and the inherent variability of ADC derived from breath hold acquisitions, respiratory triggered sequences with higher SNR may be required to detect a small difference if it exists. The application of more b values may also help detect subtle differences. A limitation of our study is the relatively small number of malignant lesions. However, the lesions showed ADC overlap with benign pathologies and while a larger series may show occasional outliers, ADC values by themselves will most likely remain poor discriminators of malignancy. Another limitation is the relatively high proportion of female patients in our cohort. This is due to MRI with hepatocellular contrast agents being used to confirm the diagnosis of FNH, which is most common in females, and was the most common solid benign lesion in our group. To our knowledge, there is no study reporting different hepatic ADC values between males and females and so this limitation is unlikely to skew our results.
Conclusion Liver MRI is a well-established tool in the characterisation of liver lesions. The application of DWI in the characterisation of solid lesions has been disappointing in the past. This study further establishes that the careful use of lesion ADC to normal liver ratios cannot distinguish solid malignant lesions from solid benign lesions.
References 1. Ferlicot S, Kobeiter H, Tran Van Nhieu J et al. MRI of atypical focal nodular hyperplasia of the liver: radiology-pathology correlation. AJR Am J Roentgenol 2004; 182: 1227–31. 2. Zech CJ, Grazioli L, Breuer J, Reiser MF, Schoenberg SO. Diagnostic performance and description of morphological features of focal nodular hyperplasia in Gd-EOB-DTPA-enhanced liver magnetic resonance imaging: results of a multicenter trial. Invest Radiol 2008; 43: 504–11. 3. Parikh T, Drew SJ, Lee VS et al. Focal liver lesion detection and characterization with diffusion-weighted MR imaging: comparison with standard breath-hold T2-weighted imaging. Radiology 2008; 246: 812–22. 4. Coenegrachts K, Delanote J, Ter Beek L et al. Improved focal liver lesion detection: comparison of single-shot diffusion-weighted echoplanar and
36
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
single-shot T2 weighted turbo spin echo techniques. Br J Radiol 2007; 80: 524–31. Bruegel M, Holzapfel K, Gaa J et al. Characterization of focal liver lesions by ADC measurements using a respiratory triggered diffusion-weighted single-shot echo-planar MR imaging technique. Eur Radiol 2008; 18: 477–85. Gourtsoyianni S, Papanikolaou N, Yarmenitis S, Maris T, Karantanas A, Gourtsoyiannis N. Respiratory gated diffusion-weighted imaging of the liver: value of apparent diffusion coefficient measurements in the differentiation between most commonly encountered benign and malignant focal liver lesions. Eur Radiol 2008; 18: 486–92. Sandrasegaran K, Akisik FM, Lin C, Tahir B, Rajan J, Aisen AM. The value of diffusion-weighted imaging in characterizing focal liver masses. Acad Radiol 2009; 16: 1208–14. Onur MR, Cicekci M, Kayali A, Poyraz AK, Kocakoc E. The role of ADC measurement in differential diagnosis of focal hepatic lesions. Eur J Radiol 2012; 81: e171–6. Moteki T, Horikoshi H, Oya N, Aoki J, Endo K. Evaluation of hepatic lesions and hepatic parenchyma using diffusion-weighted reordered turboFLASH magnetic resonance images. J Magn Reson Imaging 2002; 15: 564–72. Sun XJ, Quan XY, Huang FH, Xu YK. Quantitative evaluation of diffusion-weighted magnetic resonance imaging of focal hepatic lesions. World J Gastroenterol 2005; 11: 6535–7. Kilickesmez O, Bayramoglu S, Inci E, Cimilli T. Value of apparent diffusion coefficient measurement for discrimination of focal benign and malignant hepatic masses. J Med Imaging Radiat Oncol 2009; 53: 50–5. Demir OI, Obuz F, Sagol O, Dicle O. Contribution of diffusion-weighted MRI to the differential diagnosis of hepatic masses. Diagn Interv Radiol 2007; 13: 81–6. Agnello F, Ronot M, Valla DC, Sinkus R, Van Beers BE, Vilgrain V. High-b-value diffusion-weighted MR imaging of benign hepatocellular lesions: quantitative and qualitative analysis. Radiology 2012; 262: 511–9. Taouli B, Martin AJ, Qayyum A et al. Parallel imaging and diffusion tensor imaging for diffusion-weighted MRI of the liver: preliminary experience in healthy volunteers. AJR Am J Roentgenol 2004; 183: 677–80. Kwee TC, Takahara T, Koh DM, Nievelstein RA, Luijten PR. Comparison and reproducibility of ADC measurements in breathhold, respiratory triggered, and free-breathing diffusion-weighted MR imaging of the liver. J Magn Reson Imaging 2008; 28: 1141–8. Taouli B, Sandberg A, Stemmer A et al. Diffusion-weighted imaging of the liver: comparison of navigator triggered and breathhold acquisitions. J Magn Reson Imaging 2009; 30: 561–8.
© 2013 The Royal Australian and New Zealand College of Radiologists
DWI for liver lesion characterisation
17. Shiehmorteza M, Sirlin CB, Wolfson T et al. Effect of shot number on the calculated apparent diffusion coefficient in phantoms and in human liver in diffusion-weighted echo-planar imaging. J Magn Reson Imaging 2009; 30: 547–53. 18. Nasu K, Kuroki Y, Sekiguchi R, Kazama T, Nakajima H. Measurement of the apparent diffusion coefficient in the liver: is it a reliable index for hepatic disease diagnosis? Radiat Med 2006; 24: 438–44. 19. Hollingsworth KG, Lomas DJ. Influence of perfusion on hepatic MR diffusion measurement. NMR Biomed 2006; 19: 231–5.
© 2013 The Royal Australian and New Zealand College of Radiologists
20. Taouli B, Tolia AJ, Losada M et al. Diffusion-weighted MRI for quantification of liver fibrosis: preliminary experience. AJR Am J Roentgenol 2007; 189: 799–806. 21. Holzapfel K, Bruegel M, Eiber M et al. Characterization of small (
