Occlusion of the superior vena cava (SVC) is a complication of malignant and benign diseases that compress, occlude, or invade the SVC, subsequently directing blood flow into collateral veins. The extent and location of collateral veins are extremely variable and often involve the chest, abdomen, and pelvis (1–4). The terminology describing these veins can be confusing because of the variable terms used to describe the same vessels and the large number of potential collateral pathways.
Previous literature suggests that there are four main collateral pathways: lateral thoracic, internal mammary, azygos, and vertebral (1–3). The classification by Stanford et al. (4) used venographic data to describe four main patterns of blood flow following partial or complete occlusion of the SVC. These include type I, partial occlusion of the SVC with patency of the azygos vein; type II, near complete to complete obstruction of the SVC with patency and antegrade flow through the azygos vein and into the right atrium; type III, near complete to complete obstruction of the SVC with reversal of azygos blood flow; and type IV, complete obstruction of the SVC and one or more major caval tributaries, including the azygos system. However, no single publication comprehensively correlates all known collateral vessels and pathways. We seek to demonstrate the spectrum of venous collateral pathways due to SVC obstruction as seen on both thoracic and abdominopelvic CT and study the frequency of collateral pathway groups as compared with previous reports that were based on only venographic and/or chest CT data (2,3,5).
SUBJECTS AND METHODS
A retrospective review of our radiologic database was performed to identify all patients with SVC obstruction who underwent CT scans between 1990 and 1999. A total of 76 patients were found. Fifty-seven of these patients had only partial SVC occlusion without collateral formation and were excluded from further review. The remaining 19 patients underwent a total of 21 CT scans that were included for this cohort. The age ranged from 19 to 89 years with an average of 48.1 years. Ten were men and nine were women. Clinical and pathologic records were reviewed in all cases to determine the etiology of the SVC occlusion. All 19 patients underwent either mediastinoscopy or surgery. Therefore, all diagnoses were pathologically proven.
All patients received contrast-enhanced helical CT scans of the chest, abdomen, and pelvis on a GE 9800 HiSpeed scanner (General Electric Medical Systems, Milwaukee, WI, U.S.A.). Seven millimeter collimation was used through the liver and 10 mm collimation was used through the chest, infrahepatic abdomen, and pelvis, with a pitch of 1:1 after injection of 140–150 ml of nonionic contrast medium at 1.5–2.5 ml/s. Standard soft tissue, lung, and liver windows were photographed. Two patients had noninfused CT of the chest in addition to the contrast-enhanced CT. All scans were retrospectively reviewed by all authors, and decisions regarding the anatomic location and vessel nomenclature were reached by consensus. Four patients had superior vena cavography performed, two of which were available for review and were correlated with corresponding CT scans.
The terms used to describe collateral pathways in 16 articles (1–16) and two textbooks (17,18) were correlated to ensure uniformity in our terminology. The location and frequency of each collateral pathway and the level of the SVC obstruction were tabulated. The classification by Stanford et al. (4) was applied to each case. The extent of the SVC obstruction was classified as either SVC alone, SVC in combination with azygos occlusion, bilateral brachiocephalic occlusion with complete or incomplete SVC occlusion, or SVC occlusion with left or right brachiocephalic vein occlusion. If the occlusion level and collateral patterns did not conform to the Stanford et al. classification scheme, the case was considered unclassified and its atypical features were tabulated accordingly. Pulmonary and hepatic shunts were included as atypical features. The number of collateral pathway groups seen was analyzed with reference to the occlusion level (Student's t test).
The etiologies of SVC obstruction in this cohort are shown in Table 1. Of 19 patients, 1 had small cell carcinoma of the lung, 4 had Hodgkin disease, 2 had non-Hodgkin disease, and 6 had metastases compressing the SVC. The source of metastases included melanoma, renal cell carcinoma, rhabdomyosarcoma of the prostate, carcinoma of the colon, esophageal carcinoma, and supraglottic laryngeal carcinoma. One patient had protein S deficiency and fibrosing mediastinitis. One patient had end-stage renal disease and two others had undergone renal transplantation. One patient had mastectomy and radiotherapy. One had congestive heart failure and thalassemia.
The frequency of collaterals in this series (Table 2) differed from that of previous reports (2,3,5). The four most common collateral veins in this cohort are, in order of decreasing frequency, the azygos vein, thoracoepigastric vein, mediastinal vein, and internal mammary vein (Fig. 1). The lateral thoracic vein communicates with the thoracoepigastric vein and is referred to by some as “anterior chest wall collaterals” (Fig. 2). The terminology describing the vertebral, esophageal, mediastinal, diaphragmatic, and thoracoepigastric veins varied greatly in the past literature (Figs. 3 and 4). We therefore suggest organizing individual collateral veins by venous plexus systems (Table 3) to eliminate the confusing array of terms used to describe similar vessels in the literature. The most common abdominal collateral veins were those along the liver surface (52.3% of cases). However, 18 patients (94.7%) had at least one collateral vessel visible in the abdomen.
The classification of SVC obstruction and the collateral veins in this series is compared with the classification scheme by Stanford et al. (Table 4). In Table 4, the cases are listed in groups of same occlusion level and in an order of ascending complexity within each group. Only 8 of 21 cases (38%) can be categorized by the Stanford scheme. Seven of 21 cases (33%) demonstrated flow in all four major collateral pathways, and 8 cases showed flow in three pathways (Table 4). A total of 14 cases (62%) could not be classified by the Stanford scheme. Furthermore, seven of the eight cases that generally fit the Stanford scheme showed some additional features that varied from the classification description.
Four patients also developed systemic-portal shunts with hypervascular foci in the liver (Fig. 5). One patient with history of fibrosing mediastinitis developed a systemic-pulmonary shunt (Fig. 6). One patient studied with a noninfused CT showed extensive calcifications abutting and possibly within the SVC (Fig. 7), which might have been mistaken for contrast enhancement had only an infused study been done.
We found no statistically significant relationship between the level of occlusion and the number of collateral pathway groups seen. However, all eight cases with omental varices had either SVC obstruction alone (two cases) or in combination with bilateral brachiocephalic vein occlusion (six cases).
The two available vena cavograms correlated with corresponding CT scans and were helpful in verifying an otherwise confusing and tortuous course of the azygos vein (Fig. 8). CT showed more collateral veins than vena cavography in these patients.
CT findings in SVC occlusion include an unopacified SVC with or without opacification of the brachiocephalic veins and other collateral pathways (1–3,5,9). The presence of collateral veins, regardless of number and location shown on CT scans, is an accurate predictor of SVC syndrome, with a sensitivity of 96% and a specificity of 92%(2). The wide range in frequency of venous collateral opacification can be attributed to various factors in the contrast-enhanced CT technique: the amount and injection rate of contrast material, the timing of scanning with respect to the administration of contrast material, and possibly the positioning of the patient during scanning (2). CT and vena cavography are complementary in showing collateral pathways (1,3), as demonstrated by the two cases in this series. However, venography is usually done with a single view and is limited by superimposition of structures.
In reviewing the literature, we found a tremendous lack of uniformity in the nomenclature of collateral veins, and some terms are unique to a single publication. Sixty-two percent of our cases had collateral veins that could not be categorized using the classification scheme published by Stanford et al. (4). In contrast to other series, the most common individual collateral vein in this series is the azygos vein (90.5%), whereas other authors found the veins around the scapula, back, or shoulder to be most common (2,3).
Although omitted from the Stanford classification, we found four patients whose liver served as a shunt between the systemic and portal venous system (Fig. 5) and one patient with a systemic-pulmonary shunt (Fig. 6), which have rarely been reported in previous literature (7,10–16,19–22). It is also interesting to note that collateral veins may involve the omentum, pelvis, and perirectal areas (Fig. 3). Nearly all patients (94.7%) had at least one collateral at or below the level of the diaphragm that could be observed on abdominal CT scans. These data suggest that the presence of these abdominal collaterals, when seen on abdominopelvic CT scans, may indicate SVC obstruction. Therefore, radiologists should be cognizant of these collateral venous pathways in interpreting abdominal and pelvic CT studies and suspect undiagnosed SVC obstruction when such collaterals are encountered.
It is also interesting to note that in the previous literature, the most common etiology for SVC obstruction in patients who develop collateral veins is direct compression of the upper central venous return by a mediastinal or pulmonary parenchymal mass (3,8,9). However, with increased placement of central venous devices in the dialysis and oncologic population in more recent times, there is an increasing incidence of catheter-related SVC occlusions (1) as reflected in our patient population from a tertiary care center. Furthermore, it is possible that many cases of tumor-induced SVC obstruction result in only partial SVC occlusion without significant collateral formations. This may have also contributed to the fact that only 28.6% of this series was related to mass or adenopathy.
We recognize that an inherent limitation of this study is the complete reliance on CT to identify collateral veins. Although some authors have acknowledged that CT is better than venography in identifying the etiology of SVC obstruction (1–3), others contend that venography may be better in identifying some, but not all, collateral veins (1,3,23).
In summary, we have described a spectrum of collateral venous formations as seen by CT. This includes not only thoracic collaterals but also abdominal and pelvic pathways in SVC obstruction. In reviewing the literature, a confusing nomenclature was encountered. Furthermore, as we attempted to classify these patterns with Stanford et al.'s published scheme (4), most cases could not be categorized. Finally, we also noted that thoracic collateral veins, when present on CT, are nearly always accompanied by collateral veins in the abdomen that, although excluded from the prior classification system, may be significant as indicators for undiagnosed SVC obstruction.
1. Yedlicka JW, Schultz K, Moncada R, et al. CT findings in superior vena cava obstruction. Semin Roentgenol 1989; 24:84–90.
2. Kim HJ, Kim HS, Chung SH. CT diagnosis of superior vena cava syndrome: importance of collateral vessels. AJR 1993; 16:539–42.
3. Engel IA, Auh YH, Rubistein WA, et al. CT diagnosis of mediastinal and thoracic inlet venous obstruction. AJR 1983; 141:521–6.
4. Stanford W, Jolles H, Ell S, et al. Superior vena cava obstruction: a venographic classification. AJR 1987; 148:259–62.
5. Qanadli SD, Hajjam ME, Bruckert F, et al. Helical CT phlebography of the superior vena cava: diagnosis and evaluation of venous obstruction. AJR 1999; 172:1327–33.
6. Hutchins WW, Kirchner PT, MacMahon H. Perfusion lung scan in superior vena cava obstruction: demonstration of venous collaterals and systemic pulmonary venous shunt. AJR 1982; 138:754–6.
7. Stockberger SM, West KW, Cohen MD. Right to left shunt from systemic venous to pulmonary venous system developing after SVC obstruction. J Comput Assist Tomogr 1995; 19:312–5.
8. McFadden PM, Jammplis RW. Superior vena cava syndrome. In: Shield TW, ed. General Thoracic Surgery. Philadelphia: Williams & Wilkins, 1994:1716–23.
9. Bechtold RE, Wolfman NT, Karstaedt N, et al. Superior vena caval obstruction: detection using CT. Radiology 1985; 1157:485–7.
10. Felson B. The superior vena cava: conventional projections. Semin Roentgenol 1989; 24:91–5.
11. Maldjian PD, Obolevich AT, Cho KC. Focal enhancement of the liver on CT: a sign of SVC obstruction. J Comput Assist Tomogr 1995; 19:316–8.
12. Lee KR, Preston DF, Martin NL, et al. Angiographic documentation of systemic portal venous shunting as a cause of a liver scan “hot spot” in superior vena caval obstruction. AJR 1976; 127:637–9.
13. Tetalman MR, Kusumi R, Gaughran G, et al. Radionuclide liver spots: indicator of liver disease or a blood flow phenomenon. AJR 1978; 130:291–6.
14. Trigaux JP, Lacrosse M, Daube A. venous return by the paraumbilical and hepatic veins in case of superior vena cava obstruction. Abdom Imag
15. Ishikawa T, Clark RA, Tokuda M, et al. Focal contrast enhancement on hepatic CT in superior vena caval and brachiocephalic vein obstruction. AJR 1983; 140:337–8.
16. Holemans JA, Howlett DC, Rankin SC. Case report: superior vena cava obstruction: unusual CT findings due to venous collaterals. Clin Radiol 1997; 52:559–60.
17. Clemente CS. Anatomy of the Human Body by Henry Gray.
ed. Philadelphia: Lea & Febiger, 1985:799–848.
18. Netter FH, Colacino S. Atlas of Human Anatomy
ed. Summit: Ciba-Geigy Corp., 1994:2226–49.
19. Wilson SW. Systemic to pulmonary venous communication in the superior vena caval syndrome. AJR 1976: 127:247–9.
20. Ho HT, Horowitz AL, Ho AC. Systemic to pulmonary venous communication (right-to left shunt) in superior vena cava obstruction demonstrated by spiral CT. Br J Radiol 1999; 72:712–3.
21. Muramatsu T, Mashimo M, Miyamae T, et al. Rare collateral pathways in superior vena cava obstruction, the development of the venous shunts between systemic veins and the left heart. Clin Nucl Med 1987; 12:241–2.
22. Herbener TE, Basile V, Nakamoto D, et al. Focal liver enhancement on contrast enhanced CT scan caused by obstruction of the superior vena cava. AJR 1997; 169:250–7.
23. Parish JM, Marschke RF, Dines DE, et al. Etiologic considerations in superior vena cava syndrome. Mayo Clin Proc 1981; 56:407–13.
Keywords:© 2001 Lippincott Williams & Wilkins, Inc.
Computed tomography; Vena cavae; Vena cava, obstruction; Shunts and shunting; Computed tomography; Thorax