Clinical Radiology (1991) 44, 27-30

Comparison of Left Coronary and Laevo-Phase Pulmonary Angiograms in Detecting Left Atrial Thrombi in Rheumatic Mitral Stenosis S. SHARMA, M. V. KUMAR*, V. M. REDDYt, U. KAUL*, M. RAJANI and P. VENUGOPAL~

Departments o f Cardiovascular Radiology, *Cardiology, and ~ Cardiothoracic and Vascular Surgery, Cardiothoracic Centre, All India Institute o f Medical Sciences, New Delhi, India

We have compared the diagnostic accuracy of left coronary and laevo-phase pulmonary angiograms in detecting left atrial thrombi in 27 consecutive patients with rheumatic mitral stenosis who underwent both these procedures prior to open-heart surgery. In 10 patients, both procedures were 'positive' for thrombus, confirmed at subsequent surgery in all instances (true positive). Both procedures were negative for thrombus in 11 patients, and none of these patients showed thrombi at subsequent surgery (true negative). Left coronary angiography only was positive for thrombus in one patient in whom no thrombus was found at surgery. Laevo-phase pulmonary angiography only was positive for thrombus in five patients, two of whom had thrombus at subsequent surgery. The sensitivity, specificity and predictive accuracy of left coronary angiogram were 83.3% and 93.3% and 90.9% and that of laevo-phase pulmonary angiogram 100%, 80% and 80% respectively. Laevo-phase pulmonary angiograms showed higher sensitivity and left coronary angiograms showed higher specificity for angiographic diagnosis. However, the differences were found to be statistically insignificant. Angiography is a reliable method for detecting left atrial thrombi if both left coronary and pulmonary angiograms are performed and both procedures are positive or negative for thrombus. Sharma, S., Kumar, M.V., Reddy, V.M., Kaul, U., Rajani, M. & Venugopal, P. Clinical Radiology 44, 27-30. Comparison of Left Coronary and Laevo-Phase Pulmonary Angiograms in Detecting Left Atrial Thrombi in Rheumatic Mitral Stenosis

Thrombosis of the left atrium or its appendage is common in rheumatic mitral stenosis and is said to occur in 22% of patients at surgery (Nichols et al., 1962) and in 35.8% of patients at autopsy (Wallach et al., 1963). Most thrombi are small and occur in the vicinity of the appendage but occasionally a large thrombus may extend along the posterior wall of the left atrium (Matsuyama et al., 1983). Localization of the thrombus is important since its presence is a contraindication to percutaneous balloon mitral valvo-plasty and necessitates open-heart surgery. The need for an imaging technique with high sensitivity and specificity for detecting these thrombi is obvious.

Conventional chest radiographs and fluoroscopy are unreliable in predicting left atrial thrombi (Arendt and Cardon, 1949; Matsuyama et al., 1983). Echocardiogra- phy has also been utilized for this purpose but is associated with low sensitivity and specificity (Spangler and Okin, 1975; Schweizer et al., 1981; Shrestha et al., t983) and is especially unreliable in detecting thrombi in the appendage which is the site of thrombus formation in as many as 84% of patients (Swanson and Steinberg, 1964; Schweizer et al., 1981; Shrestha et al., 1983). Transoesophageal echocardiography can readily visua- lize the atrial appendage and may prove helpful in detecting thrombi in it, especially in patients in whom conventional two-dimensional echocardiogram is not diagnostic (Shuaile, 1988). Recently, computed tomogra- phy and scintigraphy have also shown encouraging results in this regard but these investigations are not, as

Correspondence to: Dr Sanjiv Sharma, Department of Radiodiag- nosis, All India Institute of Medical Sciences, Ansari Nagar, New Delhi, India

yet, widely available (Nair et al., 1981; Tomoda et al., 1983; Yamada et al., 1984).

Angiocardiography, therefore, remains a widely accepted investigation for detecting these thrombi. Vari- ous procedures including left atrial opacification either by direct left atrial injection or in the laevo-phase of pulmonary angiograms, and selective left coronary angio- grams have been employed (Soloff and Zatuchni, 1956; Ormond et al., 1964; Swanson and Steinberg, 1964; Lewis et al., 1965; Parker et al., 1965; Baue et al., 1968; Cipriano and Guthaner, 1978; Colman et al., 1981; Eriksson et al., 1984; Hubbard et al., 1987). Previous studies reported conflicting results regarding the reliability of various angiographic procedures in detecting thrombi; in addi- tion, these procedures have not been compared in the same patient .

We planned this study in order to compare the predictive accuracy of laevo-phase pulmonary angio- graphy and left coronary angiography in detecting left atrial thrombi in 27 consecutive patients with rheumatic mitral stenosis who underwent both procedures prior to open-heart surgery.

MATERIAL AND METHODS

DUring the last 4 years, 98 patients with rheumatic mitral stenosis (50 males, 48 females; age range 16 to 68 years, mean age 41.8 years) underwent pre-operative left coronary and/or laevo<phase pulmonary angiogra- phy prior to cardiac surgery. The indications for these procedures included clinical suspicion of left atrial

28 CLINICAL RADIOLOGY

thrombi due to the presence of paroxysmal or persistent atrial fibrillation (75 patients, 86.2%) and a history of systemic embolism (34 patients, 39%). These procedures were performed in addition to routine cardiac catheteri- zation and angiocardiography including left ventricular and aortic root injections.

Angiocardiography was performed on commercially available Polydiagnost C (Philips) or Angiotron CMP (Siemens) Units utilizing single-plane radiography and 35 mm cinefilm. All studies were performed by the percutaneous transfemoral route using the Seldinger technique; National Institute of Heart (NIH) catheters for pulmonary angiography and pre-shaped Judkin's catheters for coronary arteriography. An ionic iodinated water-soluble contrast medium (Urografin 370, Schering AG) was used for angiographic assessment. Laevo-phase pulmonary angiograms were obtained in the right anter- ior oblique view by injecting a bolus of 40 ml of contrast at 18 ml/s into the main pulmonary artery. Selective left coronary angiograms were obtained in the right and left anterior oblique views by injecting 9 ml of contrast material at 3 cm3/s using a pressure injector during each filming sequence.

All angiograms were analysed independently by two observers. The angiographic criteria for the presence of thrombus in laevo-phase pulmonary angiograms included mobile or fixed, persistent filling defects of constant size in the contrast shadow of the left atrium or its appendage, partial or complete non-visualization of the appendage, irregular outline of the left atrium or its appendage, and an atrial chamber which appeared smaller than the atrial shadow. The angiographic criteria for a positive left coronary angiogram included the presence of neovascularity seen as a bunch of small vessels arising from the coronary arteries, coursing superiorly to the region of the left atrial appendage and terminating in a network of small vascular channels. This was seen as a blush of contrast coalescing into small 'lakes'. The angiograms which showed the presence of one or a bunch of small vessels only without any evidence of pooling of contrast ('lakes') were considered as negative for throm- bus.

All patients subsequently underwent surgery. Eleven patients underwent closed mitral valvotomy based on the angiographic findings and were excluded from the study. Another 60 patients had undergone only one angio- graphic procedure preoperatively and were, therefore, excluded from further comparative analysis. The angio- graphic and surgical results in the remaining 27 patients who underwent both procedures were subjected to statis- tical analysis in the following manner:

True positives Sensitivity- True positives + False negatives x 100%

True negatives . . x 100% Specificity = True negatives ÷ False poslhves

Predictive Accuracy = True positives

True positives + False positives x 100%

The results of left coronary angiograms and laevo- phase pulmonary angiograms were compared by apply- ing the proportion test (Z).

Table 1 - Results of angiographic-surgical correlation

Both angiographic procedures done

Both angiograms positive

Thrombus at surgery

Coronary angiogram positive but I-phase pulmonary angiogram negative

Thrombus at surgery

L-phase pulmonary angiogram positive and coronary angiogram negative

Thrombus at surgery

Both angiograms negative No thrombus at surgery

27 patients

10 patients

10 patients

l patient

0 patient

5 patients

2 patients

l t patients 11 patients

Table 2 - Comparison of the diagnostic accuracy of the angiographic procedures

Angiographic Sensitivity Specificity Predictive value procedure (%) (%) (%)

Left coronary 83.3 93.3 90.9 angiography

Laevo-phase 100 80 80 pulmonary angiography

RESULTS

There were 15 males and 12 females and the ages ranged between 16 and 59 years (mean age 38.3 years). The duration of symptoms of mitral stenosis ranged between 8 months and 7 years (mean 4.3 years). All patients were receiving long term anticoagulation at the time of angiographic evaluation and subsequent surgery.

Angiographic diagnosis of left atrial thrombus was made in 16 patients and thrombi were subsequently found at surgery in 12 patients. The angiographic and surgical findings are summarized (Table 1). The sensitivity, speci- ficity and predictive accuracy of both the procedures are summarized in Table 2. The sensitivity and specificity of angiographic diagnosis was 100% if both the procedures were 'positive' or 'negative' for thrombus, in comparison, only one angiographic procedure was 'positive' for thrombus in the remaining six patients and thrombi were found at surgery in two of these patients. The anglo- graphic features compatible with the diagnosis of throm- bus in left coronary and laevo-phase pulmonary angio- grams are shown in Figs 1 and 2.

DISCUSSION

Even though angiocardiography is widely employed for detecting left atrial thrombi, no correlative study of the available angiographic procedures has been pre- viously reported. Angiographic localization of left atrial thrombi was first reported by Soloff and Zatuchni (1956) who performed intravenous angiograms in 57 patients and detected thrombi in five patients (8.8 %), confirmed at surgery or autopsy in four (80%). Subsequently, attempts were made to opacify the left atrium by direct left atrial angiography and during the laevo-phase of pulmonary angiograms (Ormond et al., 1964; Lewis et al., 1965; Parker et al., 1965; Baue et al., 1968; Erikss0n et al., 1984). The latter has become an established diagnostic pro- cedure for detecting these thrombi (Tomoda et al., 1983).

ANGIOCARDIOGRAPHY IN DETECTING LEFT ATRIAL THROMBI

Table 3 - Laevo-phase pulmonary angiography in the detection of left atrial thrombi

29

Series No. o f Positive Sensitivity Specificity Predictive accuracy patients angiogram (%) (%) (%)

Soloff and Zatuchni, 1956 57 5 - 98.1 80 Lewis et al., 1965 124 10 71.4 99 90.9 parker et al., 1965 113 8 - - 75 Baue et al., 1968 167 17 77.7 97.8 82.3 Tomoda et al., 1983 28 - 70 88 - Eriksson et al., 1984 60 8 No surgical correlation mentioned

(a)

(b) Fig. 1 - Left coronary angiogram in the right (a) and left (b) anterior oblique views showing a bunch of vessels (small arrows) arising from the atrial branch of the left circumflex artery and terminating in contrast lakes (arrowhead).

The efficacy of laevo-phase pulmonary angiograms in detecting these thrombi in the previous studies is summar- ized (Table 3). In comparison, the present study shows higher sensitivity but lower specificity and predictive accuracy.

It has been reported that non-opacification of the appendage in laevo-phase pulmonary angiograms is frequently observed but thrombi in the appendage are rarely found at subsequent surgery (Parker et al. , 1965). In one study, the appendage could not be visualized at angiography in 72% of the patients without thrombosis at

Fig. 2 - L a e v o - p h a s e pulmonary angiogram in the right anterior oblique view showing a filling defect in the left atrial cavity (arrow).

surgery (Swanson and Steinberg, 1964). We made the angiographic diagnosis of thrombus in five patients based solely on non-visualization of the appendage; thrombi were subsequently found at surgery in four of these patients. Non-visualization of the appendage at angio- grams with no surgical evidence of thrombosis in the previous studies may be related to radiographic and technical factors, including positioning of the patients, faulty injection technique and suboptimal processing of the cinefilms (Parker et al., 1965; Lewis et al., 1965; Baue et al., 1968). Baue et al. reported that superimposition of the shadow of left bronchus on the left atrial appendage can result in a false positive angiogram. This can be avoided by using the prone position which would alter the profile of visualization of the appendage and would also prevent the left main bronchus from indenting this region of the heart. False positive angiogram can also result if the thrombus gets embolized in the interval between angio- graphy and surgery. Neither of these two factors was responsible for the false positive results in the present study. A false negative angiogram can result if the thrombus forms in the interval between angiography and surgery, or if the thrombus is flat and closely adherent to the mural surface. Such a thrombus would not distort the contour of the left atrium and hence would not be seen as a filling defect in the contrast shadow of the left atrium,

Of late, selective left coronary angiograms have been increasingly utilized for detecting left atrial thrombi (Colman e t al., 1981; Hubbard e t al. , 1987). The angio- graphic criteria" for 'positive' diagnosis and the differen-

30 CLINICAL RADIOLOGY

tial diagnosis are well documented (Effler et al., 1967; Marshall et al., 1969; Standen, 1975; Soulen et al., 1977; Bochna and Falicov, 1980; Colman et al., 198 t). Whereas Bochna and Falicov found that neovascularity in the region of the left atrium was highly suggestive of the presence of thrombus, Colman and colleagues reported a 32.8% sensitivity, 98.9% specificity and 83.3% predictive accuracy for positive radiographic diagnosis. We found a 83.3% sensitivity, 93.3% specificity and 90.9% predictive accuracy for positive diagnosis. The higher diagnostic accuracy in the present study may be related to the fact that we included only those angiograms showing definite 'pooling' of contrast as 'positive' for thrombus.

There is no correlation between the size or age of the thrombus and angiographic grade of neovascularity (Colman et al., 1981). The neovascularity is probably related to the local inflammatory reaction that occurs secondary to an effort to reabsorb the thrombus. Sub- sequent partial necrosis may result in fistula formation into the left atrium (Standen, 1975). The angiographic abnormalities are observed mainly in the narrow area of implantation of the thrombus to the atrial wall (Cipriano and Guthaner, 1978). The presence of a false positive angiogram can be explained by this hypothesis since even after embolization the remaining fibrous tissue can still have a sufficient amount of these peculiar blood vessels to be visualized at angiography. This was confirmed at histology in one patient in the study by Colman and colleagues.

Comparison of the diagnostic accuracy of these two angiographic procedures when performed in the same patient has not been previously reported. The present study provides a reliable method for detecting left atrial thrombi. If both the procedures are positive or negative for thrombus, then the sensitivity and specificity of angiographic diagnosis is 100%. Laevo-phase pulmonary angiograms appear more sensitive but left coronary angiograms show a higher specificity for positive diagno- sis. However, these differences are statistically insignifi- cant (proportion test). Therefore, we recommend that both procedures be performed in all patients when there is a clinical suspicion of left atrial thrombi which need to be excluded before planning balloon mitral valvotomy or cardiac surgery.

Acknowledgements. We are grateful to Dr D. R. Sunderam for his help in statistical analysis of the data.

REFERENCES

Arendt, J & Cardon, L (1949). The diagnosis of intra-auricular thrombosis in the living. Radiology, 53, 371-378.

Baue, AE, Baum, S, Wallace, HW, Blakemore, WS & Zinsser, HF (1968). The diagnosis of left atrial thrombus by cineangiography. Archives of Surgery, 97, 976 983.

Bochna, AJ & Falicov, RE (1980). Diagnosis of intracardiac thrombi in mitral stenosis and left ventricular dysfunction. Archives o f Internal Medicine, 140, 759 762.

Cipriano, R & Guthaner, DF (1978). Organized left atrial mural thrombus demonstrated by coronary arteriography. American Heart Journal, 96, 166-169.

Colman, T, Ubago, JLM, Figueroa, A, Pomer, JL, Gallo, G, Mortera, C et al. (I 981). Coronary arteriography and atrial thrombosis in mitral valve disease. American Journal of Cardiology, 47, 973-977.

Effier, DB, Sheldon, WC, Turner, JJ & Groves, LK (1967). Coronary arteriovenous fistulae: diagnosis and surgical management. Report of 15 cases. Surgery, 61, 41-50.

Eriksson, S, Osterman, G, Asplund, K, Hagg, E, Helmers, C, Kithner, F et aL (1984). Pulmonary artery cineangiography to demonstrate cardiac thrombi in patients with cerebral infarction. Acta Neurolo- gica Scandinavica, 69, 27-33.

Hubbard, WN, Hine, AL, Rubens, M & Donaldson, RM (1987). Visualization and cardiovascular diagnosis, 13, 22-25.

Lewis, KB, Criley, JM & Ross, RS (1965). Detection of left atrial thrombus by cineangiocardiography. American Heart Journal, 70, 612-619.

Marshall, WH, Steiner, RM & Wexler, L (1969). Tumour vascularity in left atrial myxoma demonstrated by selective coronary arteriogra- phy. Radiology, 93, 815-816.

Matsuyama, S, Watabe, T, Kuribayashi, S, Ootaki, M, Suzuki, Y, Sugihara, Met al. (1983). Plain radiographic diagnosis of thrombo- sis of left atrial appendage in mitral valve disease. Radiology, 146, 15-20.

Nair, CK, Sketch, MH, Maboney, PD, Lynchi, JD, Mods, AN & Kenney, NP (1981). Detection of left atrial and ventricular thrombi by computed tomography. A preliminary report. British Heart Journal, 45, 535-541.

Nichols, HT, Blarco, G, Morse, DP, Adam, A & Baltazar, N (1962). Open mitral commissurotomy: 'experience with 200 consecutive cases. Journal of American Medical Association, 182, 268-270.

Ormond, RS, Drake, EH & Gale, NH (1964). Angiographic study of left atrium in mitral stenosis. Radiology, 83, 277-282.

Parker, BM, Friendenberg, M J, Templeton, AW & Burford, TH (1965). Pre-operative angiocardiographic diagnosis of left atrial thrombi in mitral stenosis. New England Journal of Medicine, 273, 136 140.

Schweizer, P, Bardos, P, Erbel, R, Mayer, J, Merx, W, Messner, BJ et al. (1981). Detection of left atrial thrombi by echocardiography. British Heart Journal, 45, 148 156.

Shrestha, NK, Moreno, FL, Narciso, FV, Torres, L & Callenga, HB (1983). Two dimensional echocardiographic diagnosis of left atrial thrombus in rheumatic heart disease. A ctinicopathological study. Circulation, 67, 341 347.

Shuaile, A (1988). Transoesophageal two-dimensional echocardiogra- phy and embolic stroke. Stroke, 19, 1447.

Soloff, LA & Zatuchni, J (1956). The angiocardiographic diagnosis of left atrial thrombus. Circulation, 14, 25 32.

Soulen, RL, Grollman, JH, Pagila, D & Kreuten, T (1977). Coronary neovascularity and fistula formation. A sign of mural thrombus. Circulation, 56, 663-666.

Spangler, RD & Okin, JT (1975). Echocardiographic demonstration of a left atrial thrombus. Chest, 67, 716 721.

Standen, JR (1975). "Tumour vascularity" in left atrial thrombus demonstrated by selective coronary arteriography. Radiology, 116, 549-550.

Swanson, GE & Steinberg, I (1964). Roentgen features of the atrial appendage. American Journal of Roentgenology, 91, 311-320.

Tomoda, H, Hoshiai, M, Furuya, H, Kuribayashi, S, Ootaki, M, Matsuyama, Se t al. (1983). Evaluation of intracardiac thrombus with computed tomography. American Journal o f Cardiology, 51, 843-852.

Wallach, JB, Lukash, L & Angrist, AA (1963). Interpretation of the incidence of mural thrombi in the left auricle and appendage with particular reference to mitral commissurotomy. American Heart Journal, 43, 252 254.

Yamada, M, Haki, N, Ishikawa, K, Yashima, H, Hata, S, Okkubo, Net al. (1984). Detection of left atrial thrombi in men using Indium- 131 labelled autologous platelets. British Heart Journal, 51, 298-305.