Vascular and Interventional Radiology Original Research

Vascular and Interventional Radiology Original Research Lee et al. CT of Chronic Total and Subtotal Occlusive rterial Disease Vascular and Interventional Radiology Original Research Jung Eun Lee 1 Hee Jin Park 1 So Yeon Lee 1 Eun Chul Chung 1 Myung Ho Rho 1 Jang Gyu Cha 2 Sun Joo Lee 3 Lee JE, Park HJ, Lee SY, et al. Keywords: angiography, chronic total occlusion, CT, subtotal occlusion DOI:10.2214/JR.14.14251 Received December 15, 2014; accepted after revision March 29, 2015. 1 Department of Radiology, Sungkyunkwan University School of Medicine, Kangbuk Samsung Hospital, 108 Pyung-dong, Jongno-gu, Seoul 110-746, Republic of Korea. ddress correspondence to H. J. Park (parkhiji@gmail.com). 2 Department of Radiology, Soonchunhyang University School of Medicine, ucheon Hospital, Seoul, Republic of Korea. 3 Department of Radiology, Inje University College of Medicine, usan Paik Hospital, Seoul, Republic of Korea. WE This is a web exclusive article. JR 2015; 205:W550 W555 0361 803X/15/2055 W550 merican Roentgen Ray Society Differential Diagnosis of Chronic Total Occlusive and Subtotal Occlusive Disease of the Lower Extremity rteries Using Reverse ttenuation Gradient Sign on CT ngiography OJECTIVE. The purpose of this study is to evaluate the diagnostic usefulness of the reverse attenuation gradient sign in occlusive lower extremity arterial disease through CT angiography (CT). MTERILS ND METHODS. This study sample enrolled 45 men and eight women in the chronic total occlusion group and 30 men and seven women in the subtotal occlusion group. Luminal CT attenuation (in Hounsfield units) was measured at three points from the end of the occlusion site to the first collateral vessel s insertion point. We also used Hounsfield units to measure the CT attenuation of the opposite side artery at the same level in a similar manner. We compared each value using the Mann-Whitney U test. RESULTS. The absolute value of the mean differences in the Hounsfield units among the proximal, middle, and distal portion of chronic total occlusions were higher than those of subtotal occlusions, and this result was statistically significant (p < 0.001). The mean ratios of the Hounsfield units (Hounsfield units of the stenosed lumen divided by Hounsfield units of the opposite normal lumen) of the proximal portion of chronic total occlusions were statistically significantly lower than those of subtotal occlusions. CONCLUSION. The reverse attenuation gradient sign can be applied to the lower extremity arteries and can be helpful for differential diagnosis of chronic total occlusions from subtotal occlusions using CT. L ower extremity CT angiography (CT) is a clinically useful method to assess patients with suspected lower extremity artery disease, and this technique can be particularly helpful for detecting significant stenosis [1]. Many studies using CT have indicated that arterial stenosis detection is very high (sensitivities of 91 99% and specificities of 83 99%) [2 7]. However, with CT, the exact differentiation of chronic total occlusion from subtotal occlusion is difficult because both conditions can appear as complete filling defects of the contrast-enhanced vessel lumen, which results from the partial volume effect [1]. There are differences in the prognoses of chronic total occlusion and subtotal occlusion, in addition to technical difficulties in recanalization; therefore, the differentiation between these two lesions is clinically important [2]. Conventional angiography can be used to confirm whether a complete occlusion exists, but this approach is invasive and expensive and can increase morbidity and mortality [4, 8]. Recently, a coronary artery occlusive disease study by Li et al. [9] introduced the use of the reverse attenuation gradient sign to differentiate chronic total occlusions and subtotal occlusions. They reported that, although contrast attenuation decreases gradually along the normal artery, in chronic total occlusion, there is a reverse intraluminal attenuation gradient of vessels distal to the occlusive lesions. Li et al. also reported that this sign represents the retrograde collateral flow distal to an occlusive lesion, is highly specific to chronic total occlusion, and is useful for differentiation between chronic total occlusion and subtotal occlusion. Therefore, we sought to determine whether this sign could be applied to the lower extremity arteries and could be helpful for correctly diagnosing chronic total occlusion. The purpose of our study was to use CT to evaluate the diagnostic application of the reverse attenuation gradient sign in occlusive lower extremity arterial disease. W550 JR:205, November 2015

CT of Chronic Total and Subtotal Occlusive rterial Disease Materials and Methods Case Selection This is a study of a single institution, and the study was approved by the institutional ethics review board of our hospital; the requirement for informed consent was waived because of the retrospective study design. We retrospectively evaluated consecutive patients who had severe leg pain due to arterial occlusion in the iliac or femoropopliteal arteries and who underwent CT and conventional angiography for an angioplasty or stent insertion between May 2005 and January 2014. The following exclusion criteria were used: vessel length from the occlusion site to the insertion site of collateral vessels less than 2 cm (because we measured three points at 1-cm intervals), history of bypass surgery or percutaneous stent insertion, severe arteriosclerosis obliterans with too many calcifications in the arterial wall, or chronic total occlusions in both lower extremity arteries. We included 245 patients with iliac or femoropopliteal artery occlusions according to lower extremity CT findings. We excluded 78 patients because of a short length (< 2 cm) between occlusion site and insertion site of collateral vessels, 37 patients were excluded because of a history of revascularization, 26 patients were excluded for having chronic total occlusions in both lower extremity arteries, and another 14 patients were excluded because of severe arteriosclerosis obliterans. Thus, a total of 101 lesions in 90 patients were included in our study. We grouped all of the patients into two groups chronic total occlusion and subtotal occlusion on the basis of the results of conventional angiography. The chronic total occlusion group included patients with complete interruption of the contrast-enhanced lumen, and the subtotal occlusion group included patients with more than 50%, but incomplete, stenosis of the lumen. The group designations were reconfirmed through angiography, and all patients underwent additional conventional angiography within 1 week. ngiographic examinations were evaluated by two radiologists and a vascular surgeon, and consensus was used to categorize the patients as having chronic total occlusion and subtotal occlusion. Chronic total occlusion was defined on angiography as a lack of detectable anterograde contrast flow, whereas subtotal occlusion was defined by some detectable anterograde contrast flow through the narrowed lumen. Our study sample included 45 men and eight women (mean [± SD] age, 67.6 ± 12.2 years; range, 35 88 years) in the chronic total occlusion group and 30 men and seven women (mean age, 65.5 ± 10.8 years; range, 31 86 years) in the subtotal occlusion group. CT ngiography Protocol CT was performed using a 64-MDCT scanner (rilliance Power, Philips Healthcare) from 2005 through 2014. The scan coverage extended proximally from the T12 vertebral body to the feet distally. The average scan length was between 110 and 130 cm. CT was performed after 160 ml of the Fig. 1 87-year-old man with total occlusion of right common iliac and external and internal iliac arteries and reperfusion of distal stump by inferior epigastric deep circumflex iliac arteries on lower extremity CT angiography., Volume-rendered image shows that end of total occlusion site is supplied by inferior epigastric artery as first collateral vessel (arrow)., Coronal multiplanar reformatted image shows three measurement points (red circles) with interval of about 1 cm from end of occlusion site to point of first collateral vessel insertion. C, On axial image, luminal CT attenuation in Hounsfield units was measured at three points with intervals of about 1 cm from end of occlusion site to point of first collateral vessel insertion. Hounsfield units of opposite site were also measured at three points of same level. Circles denote ROIs. Letters and numbers denote area (), length (L), mean of HU (M), standard deviation (S). C JR:205, November 2015 W551

Lee et al. contrast medium was IV injected with an automatic power injector (Nemoto, Shigeru) at a flow rate of 4 ml/s to provide a 40-second bolus duration. The time of scanning delay was equal to contrast medium transit time, which was determined on the basis of contrast material arriving in the aorta by a test bolus or automated bolus triggering. The scan time delays ranged from 13 to 35 seconds, and the mean time delay was 17 ± 8.7 seconds. high-iodine concentration nonionic contrast agent (ioversol, 370 mg I/mL; Iversense, ccuzen) was used in all patients. CT was performed using a 64 0.625 collimation, table feed of 1.015 mm/s, and a gantry rotation time of 0.5 second. The x-ray tube voltage was 120 kv, and amperage was 220 m. The weighted CT dose index was 16.1 mgy, according to the protocol. Sections were reconstructed with a 1-mm slice thickness at 1-mm intervals. The total examination, including image data reconstruction, took 10 minutes. Image nalysis The evaluation of the CT was performed by two radiologists independently at different times. Cross-sectional images perpendicular to the vessel centerline were used for measurement. To Fig. 2 67-year-old man with subtotal occlusion of right proximal external iliac artery., Volume-rendered image shows subtotal occlusion of right proximal external iliac artery, about 1.1 cm in length (arrow)., Coronal multiplanar reformatted image shows three measurement points (red circles) with interval of about 1 cm from end of subtotal occlusion site. C, On axial image, luminal CT attenuation in Hounsfield units was measured at three points with interval of about 1 cm, from end of subtotal occlusion site. Hounsfield units of opposite site were also measured at three points at same level. Circles denote ROIs. Letters and numbers denote area (), length (L), mean of HU (M), standard deviation (S). measure the vessel lumen (in Hounsfield units), we magnified the image by 200% and manually drew an ROI to include the center of the vessel lumen on cross-sectional images [4, 10] (Figs. 1 and 2). The size of the ROI was matched to the size of the lumen (smaller than the contrast agent filled lumen). We adjusted the window level appropriately on the PCS to avoid inclusion of the calcified portion of the vessel wall. The luminal CT attenuation in Hounsfield units was measured at three points with an interval of about 1 cm (mean length, 7.2 ± 5.1 mm; range, 4 12 mm) from the end of the occlusion site toward the first collateral vessel s insertion point. To differentiate collateral vessels from branching vessels, which are seen distal to the occlusion site, we evaluated the diameter and prominence of the entire length of the vessels. When we traced these vessels from the main trunk, if the diameter and prominence of the collateral vessels increased gradually, the vessels were considered to be collateral vessels; if the vessels did not meet these criteria, they were considered to be branching vessels. We also measured the CT attenuation in Hounsfield units of the opposite artery at the same level in the same manner (Figs. 1C and 2C). We used interactive window adjustment on a PCS viewing station to avoid inclusion of small calcifications. Statistical nalysis We compared the values of the chronic total occlusion with the values of the subtotal occlusion using the Mann-Whitney U test (Tables 1 3). The differences in Hounsfield units among the proximal, middle, and distal portions of each vessel were calculated and compared (Table 2). The ratio of the Hounsfield units of each portion of the lumen and the opposite normal side was also calculated and compared (Table 3). Every p value was acquired as described in the tables. Interobserver agreement of the measurement between the readers was analyzed using an intraclass correlation coefficient [11]. Statistical analyses were performed using PSW software (version 18.0, IM). Results The mean Hounsfield units values of each vessel location are summarized in Table 1. The mean Hounsfield units of the proximal portion of the chronic total occlusion were statistically significantly lower than those of the subtotal occlusion. Other locations be- C W552 JR:205, November 2015

sides the proximal portion were not associated with statistically significant differences in TLE 1: CT ttenuation Measurements of the Lumen for Each Vessel Location CT of Chronic Total and Subtotal Occlusive rterial Disease Occlusion Proximal Hounsfield units. The absolute values of the Middle b Distal c mean differences in Hounsfield units among the proximal, middle, and distal portions of Total 346 ± 75 390 ± 73 405 ± 76 411 ± 85 407 ± 81 ± 81 the chronic total occlusion were statistically 360 ± 71 390 ± 71 409 ± 75 412 ± 81 406 ± 79 401 ± 76 significantly higher than those of the subtotal occlusion (p < 0.05; Table 2). The mean Subtotal ratio of the Hounsfield units (i.e., Hounsfield units of the stenosed lumen divided by 411 ± 89 409 ± 87 405 ± 89 ± 89 398 ± 86 394 ± 89 415 ± 99 416 ± 97 416 ± 95 408 ± 96 413 ± 98 401 ± 96 Hounsfield units of the opposite normal lumen) of the proximal portion of the chronic p total occlusion was statistically significantly < 0.001 0.34 0.56 0.96 0.60 0.52 lower than that of the subtotal occlusion 0.01 0.65 0.30 0.95 0.91 0.89 (Table 3). Figures 3 and 4 show the different Hounsfield unit trends that were observed Note Except for p values, data are mean ± SD Hounsfield units. Proximal portion of the opposite normal side vessel. between chronic total occlusion and subtotal occlusion according to vessel location. Distal portion of the opposite normal side vessel. b Middle portion of the opposite normal side vessel. The intraclass correlation coefficients for measuring Hounsfield units between readers TLE 2: Mean Differences in CT ttenuation Measurements of the Lumen for Each Vessel Location were 0.86 0.97, indicating excellent in- Difference Difference Difference Difference Difference Difference terobserver agreement (Table 4). There were Occlusion Proximal a Middle b Distal c Proximal d Middle d Distal d no statistically significant differences in the mean Hounsfield units, the mean differences in Hounsfield units, or the mean ratios of Total 44 ± 39 15 ± 25 29 ± 24 4 ± 25 6 ± 28 5 ± 19 Hounsfield units according to patient sex or 30 ± 25 18 ± 27 24 ± 20 6 ± 24 5 ± 26 5 ± 13 vessel location (i.e., whether the vessel was Subtotal an external iliac, femoral, or popliteal artery; p > 0.05). There were no adverse events during or after CT examinations according to the electronic medical record system. p 5 ± 25 9 ± 17 8 ± 16 6 ± 13 7 ± 13 7 ± 10 0 ± 14 8 ± 33 3 ± 19 7 ± 14 1 ± 10 7 ± 17 Discussion ecause the CT spatial resolution is limited, differentiation between chronic total occlusion and subtotal occlusion without conventional angiography is not always possible on CT [12]. The main focus of our study was to evaluate the diagnostic application of the reverse attenuation gradient sign, which was suggested by Li et al. [9] for differentiating chronic total occlusions and subtotal occlusions in occlusive coronary arterial disease using lower extremity CT. von Erffa et al. [12] found that there was an intraluminal attenuation decrease over the postocclusive segment in complete coronary artery occlusion and explained that this is probably an effect of the somewhat less intense filling of the lumen distal to occlusions via the collateral arteries. von Erffa et al. adopted this parameter as a sign of complete occlusion of the coronary artery. Li et al. reported that, although attenuation of contrast agent decreases gradually along the normal artery and the distal portion of subtotal occlusion, in coronary artery chronic total occlusion, there is a reverse intraluminal attenuation gradient of vessels distal to the occlusive lesions. They called this phenomenon the reverse attenuation gradient sign. We found < 0.001 < 0.001 < 0.001 0.11 0.87 0.13 < 0.001 < 0.001 < 0.001 0.27 0.29 0.19 Note Except for p values, data are mean ± SD Hounsfield units. a Hounsfield units of the middle portion of the lumen minus Hounsfield units of the proximal portion of the lumen. b Hounsfield units of the distal portion of the lumen minus Hounsfield units of the middle portion of the lumen. c Hounsfield units of the distal portion of the lumen minus Hounsfield units of the proximal portion of the lumen. d Difference with the normal side vessel. TLE 3: Mean Ratio of CT ttenuation Measurements of the Lumen for Each Vessel Location Occlusion Total 0.85 ± 0.16 0.88 ± 0.13 0.97 ± 0.12 0.97 ± 0.13 1.02 ± 0.12 1.02 ± 0.11 Subtotal 0.99 ± 0.07 0.99 ± 0.11 0.98 ± 0.05 0.99 ± 0.09 0.97 ± 0.06 0.99 ± 0.09 p < 0.001 < 0.001 0.14 0.14 0.07 0.43 Note Except for p values, data are mean (± SD) ratio of Hounsfield units of each portion of the lumen to Hounsfield units of the opposite normal side vessel. TLE 4: Intraclass Correlation Coefficients of CT ttenuation Measurements etween Readers Occlusion Proximal a Middle b Distal c Total 0.86 0.94 0.96 0.98 0.94 0.86 Subtotal 0.97 0.94 0.95 0.96 0.97 0.93 ll 0.92 0.94 0.96 0.97 0.95 0.90 Note p < 0.05 for all comparisons. a Proximal portion of the opposite normal side vessel. b Middle portion of the opposite normal side vessel. c Distal portion of the opposite normal side vessel. JR:205, November 2015 W553

Lee et al. 420 414 412 390 380 370 360 350 340 330 320 a similar pattern of contrast agent attenuation in chronic total occlusion and subtotal occlusion using lower extremity CT (Figs. 3 and 4) [9]. In agreement with a previous study, normal lower extremity artery attenuation and subtotal occlusion showed small but consistent decreases as the attenuation moved from the proximal to the distal part. In contrast, in chronic total occlusion, attenuation increased gradually from the end of occlusion site to the first collateral vessel s insertion point (Tables 1 and 3). 408 406 404 402 398 396 s seen in the coronary occlusive disease study [9], in arteries with chronic total occlusion, the distal portion of the lumen could be filled by retrograde flow earlier than the proximal portion, and the consequent late filling would cause the distal portion to be brighter than the proximal portion. The differences in Hounsfield units between each portion were larger in chronic total occlusions than they were in subtotal occlusions (Table 2). We can attribute these results to the slower flows of the collateral vessels 310 394 Fig. 3 Contrast attenuation pattern in chronic total occlusion and opposite normal site., Graph shows that absolute value of Hounsfield units of chronic total occlusion increased gradually from end of occlusion site to first collateral vessel insertion site., Graph shows that, in reverse, at same level, absolute value of Hounsfield units of opposite normal arteries showed consistent decreases as it extended from proximal to distal part. Readers 1 and 2 found similar patterns. 415 405 395 390 420 415 405 395 than those of the main arterial lumen of subtotal occlusion, which could communicate directly even though they were not patent as normal vessels (Fig. 5). Not every case was characterized by the same enhanced reverse attenuation gradient sign pattern. For example, 15% (9/59) of patients showed attenuation gradient patterns similar to those of the normal or subtotal occlusion arteries, and this phenomenon could be attributed to two factors. First, the retrograde flow had already filled all segments from the collater- 385 390 Fig. 4 Contrast attenuation in subtotal occlusions compared with opposite normal site., Graph shows absolute value of Hounsfield units of subtotal occlusion., Graph shows that opposite normal arteries consistently decreased from proximal to distal part of artery, although inclination of Hounsfield units is different. Readers 1 and 2 found similar patterns. W554 JR:205, November 2015

CT of Chronic Total and Subtotal Occlusive rterial Disease al vessel insertion points when the CT scan was performed. Second, it is possible that a branching vessel, which does not have retrograde vessel filling, was mistaken for a collateral vessel. Our study had some limitations. First, we used a 64-MDCT scanner. If a wider-detector MDCT scanner had been used, the reverse attenuation gradient sign might have been more prominent than our results suggest. The second limitation was the retrospective analysis design of the study. Third, the amount of calcium deposition in the vessel walls could have influenced the measured values. However, we tried to avoid inclusion of even small calcified portions by adjusting the window on the PCS and we initially excluded patients with severe arteriosclerosis obliterans. Fourth, the possibility of asymmetric flow, which is a common problem in CT of run-off arteries, might have altered the opacification of the vessels. Fig. 5 Illustration of Hounsfield unit trends between chronic total occlusions and subtotal occlusions according to vessel location. Luminal CT attenuation in Hounsfield units of chronic total occlusion (left) was measured at three points (circles) with interval of about 1 cm from end of occlusion site to point that first collateral vessel inserted. bsolute value of Hounsfield units of chronic total occlusion increased gradually from end of occlusion site to point of insertion of first collateral vessel (increasing size of circles indicates increased density). Luminal CT attenuation in Hounsfield units of subtotal occlusion (right) was measured at three points (circles) with intervals of about 1 cm from proximal to distal parts. bsolute value of Hounsfield units of subtotal occlusion decreased consistently as it extended toward distal part (decreasing size of circles indicates decreased density). (Drawing by Jang YM) In conclusion, the reverse attenuation gradient sign implies retrograde flow of the collateral vessels distal to the obstructed lumen. The reverse attenuation gradient sign can be easily detected by simple Hounsfield unit measurement of the poststenotic lumen of the arteries, and this technique is highly reproducible. This sign can be applied to the lower extremity arteries and can be helpful for using CT to differentially diagnose chronic total occlusion from subtotal occlusion. cknowledgment We thank Yun-Min Jang for providing the illustration in Figure 5. References 1. Fleischmann D, Hallett RL, Rubin GD. CT angiography of peripheral arterial disease. J Vasc Interv Radiol 2006; 17:3 26 2. Rubin GD, Schmidt J, Logan LJ, Sofilos MC. Multi-detector row CT angiography of lower extremity arterial inflow and runoff: initial experience. Radiology 2001; 221:146 158 3. Ofer, Nitecki SS, Linn S, et al. Multidetector CT angiography of peripheral vascular disease: a prospective comparison with intraarterial digital subtraction angiography. JR 2003; 180:719 724 4. Martin ML, Tay KH, Flak, et al. Multidetector CT angiography of the aortoiliac system and lower extremities: a prospective comparison with digital subtraction angiography. JR 2003; 180:1085 1091 5. Ota H, Takase K, Igarashi K, et al. MDCT compared with digital subtraction angiography for assessment of lower extremity arterial occlusive disease: importance of reviewing cross-sectional images. JR 2004; 182:201 209 6. Catalano C, Fraioli F, Laghi, et al. Infrarenal aortic and lower-extremity arterial disease: diagnostic performance of multi-detector row CT angiography. Radiology 2004; 231:555 563 7. Edwards J, Wells IP, Roobottom C. Multidetector row CT angiography of the lower limb arteries: a prospective comparison of volume-rendered techniques and intra-arterial digital subtraction angiography. Clin Radiol 2005; 60:85 95 8. Noaparast M, Rabani, Karimian F, et al. Diagnostic accuracy of 64 multi-slice CT angiography in assessment of arterial cut-off and run-off in comparison with surgical findings. Iran J Radiol 2011; 8:89 96 9. Li M, Zhang J, Pan J, Lu Z. Obstructive coronary artery disease: reverse attenuation gradient sign at CT indicates distal retrograde flow a useful sign for differentiating chronic total occlusion from subtotal occlusion. Radiology 2013; 266:766 772 10. Gakhal MS, Sartip K. CT angiography signs of lower extremity vascular trauma. JR 2009; 193:[web]W49 W57 11. Rosner. Fundamentals of biostatistics, 6th ed. elmont, C: Duxbury Press, 2005 12. von Erffa J, Ropers D, Pflederer T, et al. Differentiation of total occlusion and high-grade stenosis in coronary CT angiography. Eur Radiol 2008; 18:2770 2775 JR:205, November 2015 W555