Prevalence and Types of Main and Right Portal Vein Branching Variations on MDCT

Transcription

1 Atasoy and Özyürek Portal Vein MDCT Hepatobiliary Imaging Original Research A C M E D E N T U R I C A L I M A G I N G AJR 2006; 187: X/06/ American Roentgen Ray Society Y Çetin Atasoy 1 Elif Özyürek O Atasoy Ç, Özyürek E Keywords: CT angiography, liver, MDCT DOI: /AJR Received May 19, 2005; accepted after revision August 8, F 1 oth authors: Department of Radiology, Ankara University Medical School, Ibn-i Sina Hospital, Sihhiye 06100, Ankara, Turkey. Address correspondence to Ç. Atasoy (atasoy@medicine.ankara.edu.tr). Prevalence and Types of Main and Right Portal Vein ranching Variations on MDCT OJECTIVE. Our objective was to investigate the prevalence of variant main and right portal vein ramification in a large group of patients. SUJECTS AND METHODS. The study group consisted of 200 patients who underwent consecutive contrast-enhanced abdominal CT examinations performed with an 8-MDCT scanner. Two observers evaluated both thin axial sections and 3D maximum-intensity-projection and volume-rendered images for branching patterns of the main and right portal veins. RESULTS. Conventional main portal vein anatomy was present in 64.5% of the patients. In 9.5% of the patients, the main portal vein trifurcated into the left portal and right anterior and posterior portal veins. In 23.5% of the patients, the main portal vein divided into a common left portal vein right anterior portal vein trunk and the right posterior portal vein. Three patients (1.5%) had miscellaneous variations. Twenty-two (16.8%) of 131 patients with conventional main portal vein branching had variant right portal vein branching, most of which was a trifurcation, followed by an abnormally proximal origin of the segment VII vein from the right portal vein. CONCLUSION. Variant main portal vein branching seems to be very frequent. Common right anterior portal vein left portal vein trunk is far more common than trifurcation. Although less frequent, variations also occur in right portal vein branching. ranching anomalies of the main portal vein (MPV) at the hepatic hilum are known to be less frequent than those of the hepatic arteries, hepatic veins, and biliary ducts [1 4]. Although many of these variations can be managed safely with new surgical techniques [3, 5, 6], some continue to be contraindications to living donor right lobectomy or they at least make surgery difficult. The most suitable portal vein anatomy for right lobe living donor liver transplantation is the presence of conventional MPV branching, in which the right anterior portal vein (RAPV) and right posterior portal vein (RPPV) originate from the right portal vein (RPV). In this situation, only one portal vein anastomosis is made between the recipient s MPV and donor s RPV. When the RAPV and RPPV originate directly from the MPV or when the RAPV arises from the left portal vein (LPV), two portal vein anastomoses are needed, increasing the risk of postoperative portal vein thrombosis [4]. If these duplicated portal branches are close to each other, reconstruction with the bifurcation of the recipient s portal vein can be performed easily. When, however, the RAPV branches from the LPV more distally or within the parenchyma, an interposed vein graft is needed for reconstruction, making transplantation a challenging and highrisk task [3]. The reported incidence of portal vein variations ranges from 0.09% to 24% [7 10]. This discrepancy may be due to the use of different sample sizes and variations in the techniques used to outline the portal anatomy. The relative frequency of a particular anomaly differs in various studies, the most frequent anomaly being a trifurcated MPV [4, 8, 11 15]. This study was conducted to determine the incidence and types of variant portal anatomy in a large series of patients undergoing abdominal MDCT. Subjects and Methods Two hundred patients undergoing abdominal MDCT examinations comprised the study group. Patients were selected consecutively. Excluded were patients with large hepatic masses that might have distorted the intrahepatic portal venous anat- 676 AJR:187, September 2006

2 Portal Vein MDCT TALE 1: Main Portal Vein ranching Patterns Type Definition 1 MPV dividing into LPV and RPV, RPV then dividing into RAPV and RPPV 2 MPV trifurcating into LPV, RAPV, and RPPV 3 MPV dividing into RPPV and a common RAPV LPV trunk Note MPV = main portal vein, LPV = left portal vein, RPV = right portal vein, RAPV = right anterior portal vein, RPPV = right posterior portal vein. TALE 2: Relative Frequencies of Main Portal Vein ranching Patterns Type No. of Patients % Miscellaneous omy and those whose images were of poor quality, mainly because of insufficient portal venous opacification and motion artifacts. Indications for the examinations were known or suspected malignant disease in 149 cases, suspected intraabdominal abscess in 13 cases, abdominal pain in 12 cases, biliary obstruction in seven cases, hepatic mass in six cases, intestinal obstruction in five cases, trauma in five cases, acute pancreatitis in two cases, and follow-up of abdominal aortic aneurysm in one case. One hundred three (51.5%) of the patients were men and boys, and 97 (48.5%) were women and girls. The age range was 9 83 years (mean age, 49.9 years). Fig year-old man with type 1 main portal vein (MPV) anatomy. Maximum-intensityprojection image, oblique coronal view, shows MPV bifurcating into left portal vein (LPV) and right portal vein (RPV), which later divides into right anterior portal vein (RAPV) and right posterior portal vein (RPPV). All examinations were performed with an 8- MDCT scanner (LightSpeed Ultra, GE Healthcare). Scan parameters were mm collimation, 35 mm/s table feed, 2.5 mm section thickness, 1.25 mm reconstruction interval, 120 kv, 350 ma, and 0.5 seconds rotation time. Portal venous phase images for evaluation of the portal venous anatomy were obtained with a scan delay of 70 seconds after IV injection of 100 ml of nonionic 350 mg I/mL contrast medium (iohexol, Omnipaque 350, GE Healthcare) at an injection rate of 3 ml/s. Axial images were transferred to a workstation (Advantage Windows 4.0, GE Healthcare), where analyses of the portal venous anatomy were done by two independent reviewers experienced in the evaluation of the hepatic vascular anatomy and in 3D image rendering. After axial images were assessed, 3D reconstructions of the relevant anatomy were performed with the maximum intensity projection and volume-rendering methods. Conventional portal anatomy, which was categorized as type 1, was accepted as the MPV bifurcating into the RPV and LPV, the RPV then dividing into the RAPV and RPPV. Any deviation from this pattern was regarded as variant anatomy. Trifurcation of the MPV into the LPV, RAPV, and RPPV was categorized as type 2, and separate origin of the RPPV from the MPV with emergence of the LPV and RAPV from a common trunk was accepted as type 3. Table 1 shows the categories of MPV branching. As suggested by Hwang et al. [16], differentiation between type 2 and type 3 was made according to the shape of the gap between the RAPV and the RPPV. If the gap was triangular, the anatomy was classified as type 2; if the gap was rectangular, as type 3. In patients with type 3 anatomy, the length of the common RAPV LPV trunk was measured from its origin to its bifurcation. Other, miscellaneous portal vein variations, in particular dominant or codominant portal supply of the left lobe arising from the right lobe veins, also were noted. We looked for portal vein branches traversing the anatomic border between the right and left lobes of the liver, that is, the middle hepatic vein and Cantlie s line drawn between the inferior vena cava and the gallbladder fossa. When we found a branch, we measured its diameter on axial images. We also searched for any deviations of RPV branching from the conventional pattern, which includes division into the RAPV and RPPV, further giving off branches that supply segments V and VIII and segments VI and VII, respectively. After independent evaluations, findings were compared, and any discrepancies were resolved with a third consensus reading. ecause we studied the portal venous anatomy of patients undergoing routine abdominal MDCT for various clinical reasons, approval by the institutional ethical committee was not required. Results Table 2 shows the relative frequencies of each type of MPV branching. Of the total of 200 patients, 131 (65.5%) had conventional portal venous anatomy (type 1) (Fig. 1). Nineteen (9.5%) of the patients had trifurcation (type 2) (Fig. 2), and 47 (23.5%) had type 3 anatomy, whereby the RAPV and the LPV shared a common trunk (Fig. 3). In patients with type 3 anatomy, the length of the common RAPV LPV trunk ranged from 2.5 to 17.3 mm (mean, 7.9 mm). There were miscellaneous variations in three patients: One had an LPV arising from the RAPV. In one patient the LPV was absent, and the RAPV had branches that supplied the left lobe. In the third patient the left lobe had a codominant portal supply, one vein from the LPV and the other from the RAPV. These two veins anastomosed in the intersegmental fissure. In each of these three patients either dominant or codominant veins of the left lobe crossed the interlobar boundary. We did not see dominant venous supply of the left lobe arising from the right veins or dominant venous supply of the right lobe originating from the LPV in other patients. There were, however, tiny portal branches traversing the border between the right and left lobes in 13 (6.5%) of the patients. The diameter of these vessels ranged from 0.5 to 2.7 mm (mean, 1.53 mm) (Table 3). Nine veins originated from the LPV and supplied segment VIII; four veins originated from the right side and supplied segment IV. One hundred nine (83.2%) of 131 patients with conventional MPV branching also had AJR:187, September

3 Atasoy and Özyürek conventional RPV branching (Fig. 4), whereas 22 (16.8%) of these patients had variant RPV branching. Sixteen (12.2%) of the patients had RPV trifurcation. In seven of these patients the RPV trifurcated into the RAPV and segment VI and segment VII veins (Fig. 5); in two patients, into the RAPV and RPPV and a common trunk that supplied branches to segments V and VI; in Fig year-old man with type 2 main portal vein (MPV) anatomy. A, Volume-rendered image, left superior view, shows MPV dividing into left portal vein (LPV), right anterior portal vein (RAPV), and right posterior portal vein (RPPV)., Axial image shows proximal parts of three veins at level of trifurcation. A two patients, into the RAPV and RPPV and a separate segment V vein; in two patients, into the RAPV and RPPV and a separate segment VI vein; and in one patient, into the RAPV and RPPV and a separate segment VIII vein. One (0.8%) of the patients had RPV quadrifurcation into the RAPV, segment V vein, segment VI vein, and segment VII vein. In five (3.8%) of the patients, a segment VII vein originated proximal to the division of the RPV, which later bifurcated into the RAPV and segment VI vein in four patients (Fig. 6) and into the RAPV and a common vein that supplied branches to segments V and VI in one patient. Table 4 shows the relative frequencies of RPV branching patterns in patients with type 1 MPV branching. A C Fig year-old woman with type 3 main portal vein (MPV) anatomy. A, Volume-rendered image, anteroinferior view, shows MPV bifurcating into right posterior portal vein (RPPV) and common trunk (RAPV LPV) that gives rise to left portal vein (LPV) and right anterior portal vein (RAPV)., Axial image shows RPPV as first branch of MPV. C, Section more craniad than shows proximal parts of RAPV and LPV just beyond bifurcation of common trunk. 678 AJR:187, September 2006

4 Portal Vein MDCT TALE 3: Diameter of Small Portal Vein ranches Traversing the Interlobar oundary Origin, Destination Diameter (mm) Right lobe, left lobe 0.5 Right lobe, left lobe 0.8 Left lobe, right lobe 1.0 Left lobe, right lobe 1.2 Left lobe, right lobe 1.3 Left lobe, right lobe 1.4 Right lobe, left lobe 1.4 Left lobe, right lobe 1.5 Left lobe, right lobe 1.6 Right lobe, left lobe 1.9 Left lobe, right lobe 2.1 Left lobe, right lobe 2.5 Left lobe, right lobe 2.7 The findings of both observers were in very high agreement for classification of the MPV and RPV branching patterns. Discordant classifications were resolved by consensus in five cases. In one of these cases type 1 MPV anatomy was classified as type 2 anatomy by one of the reviewers. In the remaining 199 (99.5%) of the patients MPV anatomy was classified in the same manner by both observers. There was, however, slightly higher discordance regarding classification of the RPV branching pattern. Findings were discrepant in four of 131 patients with type 1 MPV anatomy in whom an RPV existed. All these patients had trifurcated RPVs that were classified as conventional RPV branching by one reviewer. RPV branching classifications were in agreement for 127 (97%) of the patients. Discussion Presurgical awareness of variant portal venous anatomy is important before graft procurement in liver transplantation, hepatic tumor resection, and placement of transjugular intrahepatic portosystemic shunts and for accurate tumor localization. Previous studies have shown the prevalence of variant portal venous anatomy ranges from 0.09% to 24% [7 10]. The prevalence in our series (34.5%) was considerably higher. We believe our figure reflects the true prevalence, because our series of patients was relatively large, and we meticulously evaluated both thin axial sections and reformatted 3D images. Studies in which only thick axial CT slices were assessed revealed prevalences Fig year-old man with conventional right portal vein (RPV) branching. Volume-rendered image, posterosuperior view, shows RPV bifurcating into right anterior portal vein (RAPV) and right posterior portal vein (RPPV). MPV = main portal vein, LPV = left portal vein. between 6% and 13.8% [8, 12, 13]. It is possible that many variations have been missed on these thick axial images. Several studies in which MDCT was used showed more frequent portal vein variations, ranging from 20% to 24% [4, 9, 17]. Although the technique in those studies was similar to the one we used, the prevalence of variations was still considerably lower than our findings, possibly because of their relatively small sample sizes. In a large angiography study, Cheng et al. [18] found portal vein variations in 30% of patients, a figure similar to ours. The relative frequencies of various portal vein branching types in our study also were different than in previous studies. Unlike the majority of previous studies, in which trifurcation was the most frequent variation [4, 8, 11 15], our study showed a common RAPV LPV trunk was almost 2.5 times more common than trifurcation. This striking difference may have been the result of use of different methods of imaging the portal venous ramification. We believe that on thick axial sections a short RAPV LPV trunk, which was not infrequent in our experience, can easily be misclassified as trifurcation. Reformations in 3D are crucial for accurate visualization of the anatomy in such cases. To differentiate a short RAPV LPV trunk from trifurcation, we used the criteria described by Hwang et al. [16], who classified the anatomy as type 3 (i.e., common RAPV LPV trunk) in instances in which the gap between the RAPV and the RPPV was rectangular. Differentiation of type 3 from type 2 anatomy has several advantages: In most donors with type 2 anatomy, despite the absence of an RPV, a single portal lumen can be obtained from the RAPV and RPPV owing to their close approximation. Type 3 anatomy, however, makes surgery more complicated, because two transections of the RAPV and RPPV are needed, resulting in two portal lumens in the right lobe graft [1]. The length of the common RAPV LPV trunk has important surgical implications. When they are close to each other, two donor portal branches can be anastomosed to the recipient s portal bifurcation; that is, the donor RPPV is anastomosed to the recipient s RPV, and the donor RAPV is anastomosed to the recipient s LPV. Such Y-grafts allow simultaneous reperfusion through both donor portal branches. When the donor veins are widely spaced, however, as with a long RAPV LPV trunk, an extension-type graft may be needed for reconstruction of the donor Fig year-old girl with trifurcation of right portal vein (RPV). Volume-rendered image, superior view, shows RPV trifurcating into right anterior portal vein (RAPV), segment VII vein (SVII vein), and segment VI vein (SVI vein). MPV = main portal vein, LPV = left portal vein. AJR:187, September

5 Atasoy and Özyürek Fig year-old woman with abnormally proximal origin of segment VII vein (SVII vein). A, Volume-rendered image, superior view, shows SVII vein originating from right portal vein (RPV) proximal to its bifurcation into right anterior portal vein (RAPV) and segment VI vein. MPV = main portal vein, LPV = left portal vein, RAPV = right anterior portal vein, SVI vein = segment VI vein. and C, Axial images show abnormally proximal origin of SVII vein from RPV. shows distal part of SVII vein supplying posterosuperior part of right lobe. C shows origin of SVII vein from right portal vein. RAPV. This step can result in delayed reperfusion of a segment of the graft [6]. The average length of the RAPV LPV trunk in our series was 7.9 mm. As far as we know, no cutoff value has been reported for length of the RAPV LPV trunk above which surgeons have to use an extension graft. Unlike in right lobe transplantation, in right posterior segment procurement the presence of type 3 anatomy is TALE 4: Relative Frequencies of Right Portal Vein ranching Patterns in Patients with Type 1 Main Portal Vein ranching Type No. of Patients % Conventional Trifurcation Proximal origin of segment VII vein Quadrifurcation more desirable in donor selection. Hwang et al. [16] reported that almost no candidate with type 1 and only 3.6% of those with type 2 anatomy were suitable for right posterior graft procurement, whereas 35.2% of patients with type 3 portal veins had suitable anatomic features for this procedure. Dominant portal venous supply of segment IV arising from the RPV may contraindicate right lobectomy in liver transplantation candidates. This rare situation was found in 2% of the population in one study [14]. Other studies have shown widely varying frequencies from 0% [1, 4, 9, 17] to 32.5% [5]. We found major portal veins crossing the interlobar boundary to supply segment IV in 1.5% of patients, although surgically insignificant small veins that originated from the RPV and supplied segment IV were present in another 2% of the study group. We found no patients with dominant portal venous supply of the right liver segments originating from the LPV, which is a relative contraindication to right lobe harvesting A C [14]. Nevertheless, nine (4.5%) of the patients had small veins originating from the left portal system and supplying the right lobe. The largest of these veins was 2.7 mm in diameter. Such tiny veins do not hinder liver donation or tumor resection and are ligated during surgery. To our knowledge, variations in RPV ramification have not been described in the radiology literature. These variations are not rare, having been found in 33.5% of cases in a cadaveric dissection study [19]. We found variant RPV ramification in 22 (16.8%) of 131 patients with type 1 MPV branching. There were several types of variations, but the most common was trifurcation, which most frequently involved separate origins of segment VI and VII veins from the RPV, as was also the case in the study by Hata et al. [19]. In five patients the segment VII vein arose from the RPV proximal to its bifurcation. In three patients segment V and segment VI veins ramified from a common trunk. The latter variation can hinder identi- C 680 AJR:187, September 2006

6 Portal Vein MDCT fication of the anterior border of segment VI [19]. Preoperative awareness of variant RPV branching may be beneficial in right posterior segment harvesting and in segmental resection involving the right lobe. Lack of pathologic confirmation of the findings may be considered a limitation of our study. However, our purpose was not to show the accuracy of CT in the detection of various portal vein variations. Rather we aimed to learn the prevalence of variant portal venous anatomy in a large population by using MDCT, which has been reliably used to map portal veins in persons who are candidates for live liver donation. In conclusion, variant ramification of the MPV may be more common than previously reported. Unlike previous studies, our study showed that the frequency of type 3 anatomy (i.e., a common RAPV LPV trunk) far exceeds that of trifurcation. The RPV also seems to have a considerable rate of variant branching with several different patterns, some of which may influence decision making regarding right lobe surgery. References 1. Varotti G, Gondolesi GE, Goldman J, et al. Anatomic variations in right liver living donors. J Am Coll Surg 2004; 198: Lee VS, Morgan GR, Lin JC, et al. Liver transplant donor candidates: associations between vascular and biliary anatomic variants. Liver Transpl 2004; 10: Nakamura TN, Tanaka K, Kiuchi T, et al. Anatomical variations and surgical strategies in right lobe living donor liver transplantation: lessons from 120 cases. Transplantation 2002; 73: Kamel IR, Kruskal J, Pomfret EA, Keogan MT, Warmbrand G, Raptopoulos V. Impact of multidetector CT on donor selection and surgical planning before living adult right lobe liver transplantation. AJR 2001; 176: Marcos A, Ham JM, Fisher RA, Olzinski AT, Posner MP. Surgical management of anatomical variations of the right lobe in living donor liver transplantation. Ann Surg 2000; 231: Marcos A, Orloff M, Mieles L, Olzinski A, Sitzmann J. Reconstruction of double hepatic arterial and portal venous branches for right-lobe living donor liver transplantation. Liver Transpl 2001; 7: Fraser-Hill MA, Atri M, ret PM, Aldis AE, Illescas FF, Herschorn SD. Intrahepatic portal venous system: variations demonstrated with duplex and color Doppler US. Radiology 1990; 177: Soyer P, luemke DA, Choti MA, Fishman EK. Variations in the intrahepatic portions of the hepatic and portal veins: findings on helical CT scans during arterial portography. AJR 1995; 164: Erbay N, Raptopoulos V, Pomfret EA, Kamel IR, Kruskal J. Living donor liver transplantation in adults: vascular variants important in surgical planning for donors and recipients. AJR 2003; 181: Lee SG, Hwang S, Kim KH, et al. Approach to anatomic variations of the graft portal vein in right lobe living-donor liver transplantation. Transplantation 2003; 75: Atri M, ret PM, Fraser-Hill MA. Intrahepatic portal venous variation: prevalence with US. Radiology 1992; 184: aba Y, Hokotate H, Nishi H, Inoue H, Nakajo M. Intrahepatic portal venous variations: demonstration by helical CT during arterial portography. J Comput Assist Tomogr 2000; 24: Akgul E, Inal M, Soyupak S, inokay F, Aksungur E, Oguz M. Portal venous variations: prevalence with contrast-enhanced helical CT. Acta Radiol 2002; 43: Guiney MJ, Kruskal J, Sosna J, Hanto DW, Goldberg SN, Raptopoulos V. Multi-detector row CT of relevant vascular anatomy of the surgical plane in split-liver transplantation. Radiology 2003; 229: Carr JC, Nemcek AA, Abecassis M, et al. Preoperative evaluation of the entire hepatic vasculature in living liver donors with use of contrast-enhanced MR angiography and true fast imaging with steady-state precession. J Vasc Interv Radiol 2003; 14: Hwang S, Lee SG, Lee YJ, et al. Donor selection for procurement of right posterior segment graft in living donor liver transplantation. Liver Transpl 2004; 10: Schroeder T, Nadalin S, Stattaus J, Debatin JF, Malago M, Ruehm SG. Potential living liver donors: evaluation with an all-in-one protocol with multi-detector row CT. Radiology 2002; 224: Cheng YF, Huang TL, Chen CL, et al. Anatomic dissociation between the intrahepatic bile duct and portal vein: risk factors for left hepatectomy. World J Surg 1997; 21: Hata F, Hirata K, Murakami G, Mukaiya M. Identification of segments VI and VII of the liver based on the ramification patterns of the intrahepatic portal and hepatic veins. Clin Anat 1999; 12: AJR:187, September