U.S. Army Medical Department, Office of Medical History
Skip Navigation, go to content

HISTORY OF THE OFFICE OF MEDICAL HISTORY

AMEDD BIOGRAPHIES

AMEDD CORPS HISTORY

BOOKS AND DOCUMENTS

HISTORICAL ART WORK & IMAGES

MEDICAL MEMOIRS

AMEDD MEDAL OF HONOR RECIPIENTS External Link, Opens in New Window

ORGANIZATIONAL HISTORIES

THE SURGEONS GENERAL

ANNUAL REPORTS OF THE SURGEON GENERAL

AMEDD UNIT PATCHES AND LINEAGE

THE AMEDD HISTORIAN NEWSLETTER

Chapter VI

Contents

CHAPTER VI

Arterial Aneurysms and Arteriovenous Fistulas

Alterations in the Cardiac Size in Arteriovenous Fistulas

Harris B. Shumacker, Jr., M. D.

Although numerous instances of cardiac enlargement associated with the presence of arteriovenous fistulas were recorded prior to the outbreak of World War II, these reports, for the most part, concerned only individual cases or small groups of cases. No series large enough to supply statistical data were on record and such conclusions as had been arrived at concerning this phenomenon were based, therefore, on clinical impressions. The experience in World War II altered this situation. At the vascular center of Mayo General Hospital detailed studies of the heart size before and after operative removal of the traumatic arteriovenous fistula were carried out on a large number of patients with this type of lesion. These studies have made possible certain conclusions based on factual data.

MATERIALS AND METHODS

The studies reported in this chapter were carried out on 185 soldiers with traumatic arteriovenous fistulas which in all but a few instances followed combat-incurred injuries. Only 1 patient was a woman and most of the group were young adults.

  Clinical Manifestations. This series does not include any patient in whom organic heart disease was known to exist. While a number of patients complained of dyspnea on exertion, palpitation, or a distressing pounding of the heart especially when recumbent, many others who presented roentgenologic evidence of a considerable increase in cardiac frontal area complained of no symptoms which were referrable to the heart.  Among the 185 soldiers there were only 2 instances of frank cardiac failure. In 1 the episode occurred before the first roentgenologic examination was made. This patient had 2 traumatic fistulas, 1 involving the external iliac vessels and 1 involving the hypogastric vessel. The severe dyspnea, orthopnea, and edema originally present after his injury disappeared after resection of the iliac fistula in a hospital overseas. Although his heart was still enlarged he was relatively asymptomatic between the first operation and the second, when excision of the hypogastric fistula brought about a further decrease in the cardiac frontal area.


207

The second patient with frank cardiac failure, whose cardiac frontal area had. progressively increased in size, became extremely orthopneic when he experienced a small pulmonary infarction and a Streptococcus viridans septicemia as the result of infection of the fistula (see Chapter IV).  After the fistula had been excised he was relieved of all symptoms and the heart returned to normal size.

  Methods.  Teleoroentgenograms were made in all 185 patients in the series, usually both before and after operation. The predicted and actual frontal areas of the cardiac silhouette were calculated according to the method of Ungerleider and Gubner,1 calculations of the predicted values being made according to the height and weight of the patients on their admission to the vascular center.

Before operation, whenever the location of the fistula permitted it, several determinations of the blood pressure and pulse, with and without occlusion of the fistula, were made. The precise findings at operation were recorded in each patient; in most instances these observations included actual measurements or a careful estimation of the size of the fistula.

ANALYSIS OF DATA

  Regional Distribution.  An analysis of the 185 patients with arteriovenous fistulas from the standpoint of location of the lesion (Table 17), showed that in 131 these involved vessels of the pelvis and lower extremity, while in 54, vessels of the head, neck, and upper extremity were affected. (There were 53 patients with fistulas of the femoral artery-the highest number for any single vessel.)

Age Distribution. The age range in this series was 19 to 46 years; average age was 24.5 years. The age range varied somewhat in the different anatomic groups (Table 17), but average ages for patients with lesions of the pelvis and lower extremity (24 years) and for lesions of the head, neck, and upper extremity (25.7 years) were not widely different. The lowest average age, 22.4, occurred in patients with lesions of the femoral artery and the highest, 27.7 years, in those with lesions in the miscellaneous group of vessels in the upper part of the body not segregated into special categories.

Duration of Fistula. The time interval between injury and operation in these 185 patients varied from 1.0 to 30.0 months and averaged 5.4 months (Table 18). However, there was little difference in the average duration of the fistula in most of the regional categories. The highest average duration, 9.2 months, was in the category which included the profunda femoris, hypogastric, obturator, and superior gluteal arteries and can probably be explained by the small number of cases (9) in this group. The data on the time interval between injury and the final preoperative roentgenogram and between operation and the final postoperative roentgenogram are given in Table 18 and require no special comments.

1 Ungerleider, H. E., and Gubner, R.:  Evaluation of heart size measurements.  Am. Heart J. 24: 494-510, Oct 1942.


208

TABLE 17. LOCATION OF LESION AND AGE DISTRIBUTION

TABLE 18.   INTERVAL IN MONTHS BETWEEN INJURY AND PREOPERATIVE ROENTGENOGRAM, BETWEEN INJURY AND OPERATION, AND BETWEEN OPERATION AND POSTOPERATIVE ROENTGENOGRAM


209

MEASUREMENTS OF THE CARDIAC FRONTAL AREA

Preoperative measurements of the cardiac frontal area were made in 153 of the 185 patients in this series, postoperative measurements were made in 161, and both preoperative and postoperative measurements in 132 (Tables 19, 20, and 21).

TABLE 19. PROPORTIONATE DISTRIBUTION OF PREOPERATIVE MEASUREMENTS OF THE CARDIAC FRONTAL AREA

Control Measurements.  It proved impossible because of the heavy workload in this vascular center to take measurements of the cardiac frontal area in a comparable group of soldiers without arteriovenous fistulas but with otherwise similar physical and nutritional states. In the true sense of the term, therefore, this series provides no data of a control group. An analysis was made, however, of the final postoperative cardiac measurements in 119 patients whose fistulas had been removed by operation and whose preoperative measurements had not exceeded 115 percent of normal (Chart 25). Excluded from this group were patients in whom there was a great increase in heart size before operation, in whom it seemed likely that hypertrophy as well as dilatation of the heart might have occurred, and in whom some degree of cardiac enlargement might have persisted after operative removal of the fistula.


210

TABLE 20.  PROPORTIONATE DISTRIBUTION OF POSTOPERATIVE MEASUREMENTS OF CARDIAC FRONTAL AREA

TABLE 21. PROPORTIONATE DISTRIBUTION OF POSTOPERATIVE ALTERATIONS IN CARDIAC FRONTAL AREA


211

Chart 25. Distribution of measurements of cardiac frontal area (in terms of predicted area) after operative cure of arteriovenous fistulas in 119 patients in whom the preoperative measurements had not exceeded 115 percent of the predicted. Note that the majority of the values fall within the 85-105 percent range.

In these 119 patients (Chart 25) the postoperative measurements of the cardiac frontal area tended to center around 95 percent of the calculated predicted values. Almost three-fourths of the patients (74.8 percent) had measurements ranging from 86 to 105 percent of predicted values.  In 11.8 percent actual measurements were less than 85 percent of predicted values, and in 13.4 percent they were higher than 105 percent of these values. When 100 percent of the predicted value was used as the presumed normal figure, a smaller number of patients (69.8 percent) fell into the + 10 percent range,


212

and the measurements in the remaining patients were less evenly distributed. Only 5.0 percent of the patients then had values in excess of 110 percent while 25.2 percent had values lower than 90 percent of the predicted values.

Comparative Preoperative-Postoperative Measurements of the Cardiac Frontal Area

  Preoperative measurements were made on 153 patients (Table 19, Charts 26 and 27). Of these, 54.9 percent had measurements of the cardiac frontal area in excess of 105 percent of the predicted values; in 11.8 percent they exceeded 125 percent of these values. Forty-one and eight-tenths percent had measurements between 85 and 105 percent of the predicted size, while 3.3 percent showed measurements less than 85 percent of this size.  Postoperative measurements of the cardiac frontal area were made on 161 patients (Table 20, Charts 26 and 27). In contrast to preoperative values, postoperative measurements were in excess of 105 percent of the predicted size in only 26.7 percent of those tested, while in no instances were these measurements in excess of 125 percent. After operation almost two-thirds of the patients (64.6 percent) had measurements in the 85-105 percent of the predicted range, while 8.7 percent had measurements less than 85 percent of the predicted size.

  On 132 patients both preoperative and postoperative measurements were made (Table 21, Chart 28). In 37.9 percent there was no essential difference in the postoperative as compared to the preoperative values. In 3.8 percent a slight increase in cardiac size was recorded after operation. Nearly 60 percent of the patients, however, showed a reduction of 6 percent or more in cardiac size after operation, while in 9.8 percent of the patients the decrease was 21 percent or more.

Influence of Special Factors on Cardiac Measurements

  When the preoperative and postoperative measurements of the cardiac frontal area were compared in special categories of fistulas, wide differences immediately became apparent (Tables 19, 20, and 21, Chart 29). As a rule, patients with femoral and popliteal fistulas tended to show greater increases in the cardiac size before operation than those with fistulas in other locations, and patients with fistulas of the pelvis and lower extremities tended to show cardiac enlargements more frequently than those with fistulas in the upper part of the body. A similar, but less pronounced, difference was apparent when these two groups were compared with respect to the degree of reduction in the cardiac size after operation.

  Further analysis of the data suggested a direct relationship between the size of the fistula and the degree of cardiac enlargement. Fistulas in the vessels of the leg, for instance, which were generally small as compared with fistulas of the femoral and popliteal vessels, showed less cardiac enlargement than the group with femoral and popliteal lesions. In order to evaluate this


213

Chart 26.  Distribution of measurements of cardiac frontal area (in terms of predicted area) in 158 patients before and in 161 patients after operative cure of arteriovenous fistula.

observation more exactly, the degree of cardiac enlargement was analyzed in 36 femoral and 31 popliteal arteriovenous fistulas of approximately the same duration and in which the size was known (Table 22). It was evident that those with larger fistulas tended to show more cardiac enlargement. When popliteal and femoral fistulas of roughly the same size were compared, it seemed evident that the femoral fistulas were associated with more cardiac enlargement; this appeared logical since the femoral arteries were larger in diameter and their fistulas closer to the heart. Indeed, the increase in cardiac frontal area was as great in the smaller femoral fistulas as in the larger popliteal fistulas.

  A similar comparison of the preoperative cardiac measurements in a small number (34) of subclavian, axillary, popliteal, and femoral arteriovenous fistulas of approximately the same diameter (7 mm. or more) and approximately the same duration (from 3.1 to 6.5 months) supported the evidence already adduced for femoral and popliteal arteriovenous fistulas (Table 23). Although the numbers are small, the analysis indicates that femoral fistulas


214

Chart 27. Proportionate distribution of measurements of cardiac frontal area in percentage of predicted area in 153 patients before and in 161 patients after operative cure of arteriovenous fistula.

TABLE 22. PROPORTIONATE DISTRIBUTION OF CARDIAC ENLARGEMENT IN RELATION TO SIZE OF FISTULA IN PATIENTS WITH POPLITEAL AND FEMORAL ARTERIOVENOUS FISTULAS


215

Chart 28. Proportionate distribution of alteration in cardiac frontal area in 18,2 patients after operative cure of arteriovenous fistula.

are likely to be associated with the greatest increases in cardiac frontal area, and subclavian and axillary fistulas with the least. The popliteal, axillary, and subclavian vessels were roughly equal in size; the femoral vessels were larger. On the other hand, the subclavian and axillary arteriovenous fistulas were closest to the heart and the popliteal fistulas the most distant.

An analysis of the possible relationship of the duration and size of the fistula to the preoperative measurements of the cardiac frontal area (Table 24) revealed that in 31 popliteal fistulas in which the time element was essentially the same, the size of the heart could be correlated with the size of the fistula; smaller fistulas were associated with smaller heart size and larger fistulas with larger heart size. Thirty-six femoral arteriovenous fistulas were similarly studied. The fistulas associated with the greatest increase in cardiac frontal area were larger fistulas of longer duration than those associated with less cardiac enlargement.


216

TABLE 23. PROPORTIONATE DISTRIBUTION OF PREOPERATIVE CARDIAC FRONTAL AREA IN VARIOUSLY LOCATED FISTULAS 7 MILLIMETERS OR MORE IN DIAMETER AND FROM 3.1 TO 6.5 MONTHS DURATION

Chart 29. Comparative distribution of preoperative measurements of cardiac frontal area (in percentages of predicted values) and of postoperative decreases (percentages of preoperative values) in fistulas caudad and cephalad to the heart. The preoperative values for the fistulas of the pelvis and lower extremities caudad to the heart are based upon 108 patients and for the fistulas of the head, neck, and upper extremities cephalad to the heart in 45 patients. The postoperative values are based, for the corresponding fistulas, upon 99 and 33 patients, respectively.


217

When 22 popliteal and 28 femoral arteriovenous fistulas of large size (7 mm. or more in diameter) were analyzed in relation to the duration of the lesion and the increase in the measurements of the cardiac frontal area (Table 25), some relationship between the two variables appeared to exist: The group of patients in whom the greatest increases in cardiac size had occurred included the majority of cases of relatively longest duration.

TABLE 24.  RELATIONSHIP OF DURATION AND SIZE OF FISTULA TO PREOPERATIVE CARDIAC FRONTAL AREA

TABLE 25. RELATIONSHIP OF DURATION OF FISTULA TO PREOPERATIVE CARDIAC FRONTAL AREA IN PATIENTS WITH POPLITEAL AND FEMORAL FISTULAS OF LARGE SIZE (DIAMETER OF 7 MM.+)

Serial preoperative determinations were made in 39 patients in this series (Table 26). In 18 the cardiac frontal area was the same at the first and last examinations and in 5 the changes in terms of the predicted size were less than 5 percent in one direction or the other.  In 4 other patients the heart size on the last examination had decreased by from 5 to 11 percent and in the remaining 12 it had increased by from 5 to 33 percent. It may be of interest to note that in 7 of the 14 patients in which the interval between examinations was 6 weeks or more, an increase in the size of the heart was noted on the second examination.

  Serial postoperative determinations of the heart size were made in 24 patients (Table 27). In 15 there was no essential change between the first and last examination; in 1 the heart had increased in size, and in 8 there was a


218

decrease of 5 percent or more in cardiac size between the first and last postoperative examinations. In approximately two-thirds of the patients the interval between examinations was a month or more.

TABLE 26.  COMPARISON OF CARDIAC FRONTAL AREA IN PATIENTS IN WHOM MORE THAN ONE MEASUREMENT WAS MADE BEFORE OPERATION


219

TABLE 27.  COMPARISON OF CARDIAC FRONTAL AREA IN PATIENTS IN WHOM MORE THAN ONE MEASUREMENT WAS MADE AFTER OPERATION

Finally, the response of the pulse, systolic and diastolic blood pressure, and pulse pressure to temporary occlusion of the fistula was correlated with the frontal cardiac measurements in terms of the predicted heart size (Chart 30). Only 20 to 30 percent of the patients who showed a minimal or moderate response to occlusion of the fistula had distinct degrees of cardiac enlargement, while between 47 and 69 percent of those who showed maximal response to it also showed maximal increases in the size of the heart.

COMMENT

Although it is now well recognized that arteriovenous fistulas are likely to cause an increase in the cardiac output, enlargement of the heart, and in some instances cardiac failure, it is interesting to note that these potentialities have been appreciated only during the last 25 years.


220

Chart 30.  Comparative measurements of cardiac frontal area (in percentages of predicted value) in relation to magnitude of response of pulse and blood pressure to temporary occlusion of the fistula. The measurements are based upon 145 patients.

In a discussion of arteriovenous fistulas in 1915, Osler 2 wrote:

. . . we all agree, I think, with the conclusion arrived at by Soubbotitich, senior surgeon of the Belgrade State Hospital, from his experience in the Balkan War, that arteriovenous aneurysms should be operated upon, as they offer small prospect of spontaneous cure, although they often remain stationary for a long time and cause relatively little trouble.'


2 Osler, Sir W.:  Remarks on arterio-venous aneurysm. Lancet 1: 949-955, 8 May 15.


221

Osler's convictions on the matter, however, must have been lukewarm at best, since in reference to one of his own patients he wrote:

  The tumor had increased, and the question was whether it was safe to leave him alone. This was the policy I urged strongly. Twice he narrowly escaped operation. [Italics not in original.]

When he enumerated the end results of arteriovenous fistulas Osler did not mention the possibility of cardiac strain and failure, although later in the same paper he did state:

  Remote effects on the general circulation are rare, particularly in aneurysms of the vessels of the head and arms. One of my patients . . . died from heart disease which may have had some connection with his long-standing lesion.

As Reid 3 wrote in 1925:

. . . it has taken the profession a long time to establish a cause and effect between arteriovenous fistulas, in vessels smaller than the aorta, and the cardiac disturbances they produce.  So often we remain blind to the insidious conditions that shorten the duration of human lives. Although Osler for years followed two cases of arteriovenous aneurysms (one axillary, the other femoral) in which the patients died from cardiac disease at the early ages of 29 and 46, he was never, unless possibly shortly before his death, convinced of the causal relationship existing between the two conditions; and for a long time he counselled against operations on arteriovenous aneurysms.

  Since Osler was the acknowledged leader of his profession, there can be little doubt, as Reid intimates, that the best medical opinion of his day was in agreement with his point of view.

  Between 1920 and 1924, Reid,4 Matas,5  and Holman 6 all pointed out the profound effects of arteriovenous fistulas upon the heart. Their clinical experience was necessarily limited by the infrequency of these lesions, but their observations were confirmed by the important experimental studies of Reid 7 and Holman.8 Since that time a number of instances of cardiac enlargement and cardiac failure in association with arteriovenous fistulas have been placed on record, particularly in recent years. Though most of the instances of frank cardiac failure have occurred in patients with large arteriovenous fistulas of long duration, a number have been recorded in which failure occurred soon after the appearance of the fistula. Mason,9 for example, reported a remarkable case in which failure occurred within a few weeks after the appearance of a traumatic subclavian arteriovenous fistula.

3 Reid, M. R.:  Abnormal arteriovenous communications, acquired and congenital; effects of abnormal arteriovenous communications on the heart, blood vessels and other structures.  Arch. Surg. 11: 25-42, Jul 1925.

4 Reid, M. R.:  Effect of arteriovenous fistula upon the heart and blood vessels.  Bull. Johns Hopkins Hosp. 31: 43-50, Feb 1920.

5 Matas, R.:  On the systemic or cardiovascular effects of arteriovenous fistulae; a general discussion based upon the author's surgical experience.  Tr. South. S. A. 36: 623-681, 1923.

6 (1) Holman, E.:  The physiology of an arteriovenous fistula.  Arch. Surg. 7: 64-82, Jul 1923.

(2) Holman, E.:  Experimental studies in arteriovenous fistulas; cardiac dilatation and blood vessel changes. Arch. Surg. 9: 856-879, pt. 2, Nov 1924.

See footnote 4, above.

See footnote 6, above.

Mason, J. M.:  Extreme cardiac decompensation following traumatic arteriovenous fistula of left subclavian vessels. Am. J. Surg. 20: 451-473, May 1933.


222

Because most recorded cases of arteriovenous fistulas concern individuals or small groups, the literature supplies little information concerning the incidence of cardiac enlargement in patients with this condition. Holman 10 noted that cardiac dilatation and hypertrophy, although they had been mentioned in only 4 percent of the 447 cases collected by Callander,11 had been present in 6 of the 21 cases which he himself reviewed from the Johns Hopkins Hospital.

Postoperative decreases in the cardiac measurements have also been recorded. Pendergrass12 examined 32 patients in whom arteriovenous fistulas had been removed surgically and found that 27 showed an average decrease of 1.18 cm. in the cardiac diameter. One of the 5 remaining patients had an increase in cardiac size after operation and in the other 4 patients there was no essential change.

The method proposed by Ungerleider and Gubner 13 for measuring the frontal area of the heart has proved as reliable as any other method devised for this purpose and was used in this study. In calculating the predicted frontal area, the height and weight of the patient on admittance to the hospital were employed and the values so calculated were used in each patient for determining the percent of the predicted value in all subsequent examinations. Had the predicted value been recalculated at each subsequent examination, apparent decreases in the cardiac frontal area would probably have been observed in most patients even if no change had actually occurred because most of the patients gained weight after they had been admitted to the vascular center, and, according to the tables for estimation, the predicted frontal area for a patient of any given height increases as the weight increases.

  It is unfortunate that no "normal" values in a similar group of young soldiers without fistulas and in a comparable state of nutrition were available for study. The tables used for determination of the predicted cardiac frontal area use 100 percent as normal, and it is generally accepted that values which differ by more than 10 percent from those predicted should be regarded as abnormal. Had this 100 percent been used as normal in this series, a large number of the postoperative measurements would have seemed less than normal. An analysis of the final postoperative measurements of the cardiac frontal area in 119 patients cured of arteriovenous fistulas showed that they did indeed tend to center around 95 percent of the predicted values rather than 100 percent. Those which were beyond the limits of +10 percent of the control values were evenly distributed, about one-half above and about one-half below these levels. The 119 patients used for analysis did not include any in

10  See footnote 6 (1), p. 221.

11  Callander, C. L.:  Study of arteriovenous fistula with an analysis of 447 cases.  Ann. Surg. 71: 428-459, Apr 1920.

12 Pendergrass, R. C.:  Cardiac changes in arteriovenous fistula.  Am. J. Roentgenol. 53: 423-431, May 1945.

13 See footnote 1, p. 207.


223

which there was marked preoperative enlargement since it was felt that in such patients hypertrophy as well as dilatation might have occurred, with some degree of enlargement persisting after operation.

In these studies from the vascular center of Mayo General Hospital evidence of enlargement of the heart was demonstrable in about 50 percent of the 185 young patients with peripheral arteriovenous fistulas of relatively short duration. This fact was confirmed by evidence of measurable reduction in cardiac size after operation in a comparable percentage.

The data derived from this investigation tend to confirm certain opinions, previously expressed, as to factors important in the development of cardiac enlargement in the presence of arteriovenous fistulas. These opinions, al though based primarily upon experimental work, were supplemented, at least in part, by clinical observations. The conclusions drawn at this vascular center may be stated as follows:

  1.  In patients with arteriovenous fistulas some relationship appeared to exist between the degree of cardiac enlargement and the size of the fistula, particularly when it was located in vessels of the pelvis and lower extremities. It should be noted, however, that the number of patients involved in this study was small.

2. A relationship was also evident between the duration of the fistula and the degree of cardiac enlargement. When appropriate samples of femoral and popliteal fistulas were used for comparison, with variable factors properly controlled, it was evident that cardiac enlargement varied directly with the duration of the lesion. Moreover, when roentgenograms were made in a series at intervals of 6 weeks or more before operation, 50 percent of the patients in this group showed on the last examination a demonstrable increase in cardiac size as compared with the earlier examination.

3.  The data in this series concerning changes in the size of the heart in association with femoral and popliteal arteriovenous fistulas tend to support the opinion, rather generally held, that the nearer the fistula is to the heart, the greater is the likelihood of early cardiac enlargement. Not too much importance, however, can be attached to this relationship since the caliber of the involved vessels may also have played a part. Certain of the experimental work cited to support the thesis that the distance of the fistula from the heart has an important influence on the associated cardiac enlargement can be challenged for the same reason, that is, failure to take into consideration the size of the artery selected for communication with the vein. As a rule, the closer to the heart the fistula was created in these experimental studies, the larger was the artery selected for the communication. It is apparent that if the diameter of the fistula exceeds that of the parent artery, the blood flow through the fistula will be limited in part by the size of that artery.


224

4.  These data also demonstrated that another consideration was more important in the production of cardiac enlargement than the proximity of the fistula to the heart, namely, whether it was cephalad or caudad to the heart. Both the frequency and degree of cardiac enlargement were greater in fistulas of the pelvis and lower extremities than in fistulas of the head, neck, and upper extremities. In this connection, it is curious that in years past a number of authors, although they had only a limited number of patients on whom to base the observation, made the same point. Thus Osler,14 who, as previously pointed out, was not impressed by the possible relationship between arteriovenous fistulas and cardiac strain, stated that remote effects upon the general circulation were extremely uncommon, "particularly in aneurysms of the vessels of the head and arms."

  The data in this study seem to permit two clinical conclusions. The first is that a marked change in the pulse or blood pressure upon temporary occlusion of an arteriovenous fistula is likely to be associated with an early and significant increase in the frontal area of the heart. The second is that without other information it is possible to gain some idea of the likelihood of early cardiac enlargement from the location of the fistula alone. Though a considerable increase in the size of the heart may occur relatively soon after the development of an arteriovenous fistula involving any reasonably large vessel in any portion of the body, the alteration is particularly likely to be associated with fistulas in the great vessels caudad to the heart.

14 See footnote 2, p. 220.