Variables that affect the expansion outcome of small abdominal aortic

Transcription

1 Variables that affect the expansion outcome of small abdominal aortic Jack L. Cronenwett, MD, Steven K. Sargent, MD, Michael H. Wall, MD, Mary L. Hawkes, RN, Daniel H. Freeman, Phi3, Bradley J. Dain, MS, Joel K. CurE, MD, Daniel B. Walsh, MD, Robert M. Zwolak, MD, PhD, Martha D. McDaniel, MD, and Joseph R. Schneider, MD, Hanover, N.H. rate and aneurysms Seventy-three patients with small (<6 cm in diameter) abdominal aortic aneurysms (AAAs) were selected for nonoperative management and followed up with sequential ultrasound size measurements. Fifty-four men and 19 women, 51 to 89 years of age (mean 70 years), had an initial mean AAA size of 4.1 cm (anteroposterior) 4.3 cm (lateral) diameter, with a calculated elliptic cross-sectional area of 14.3 cm 2. After a mean of 37 months of follow-up, AAA area increased at a mean rate of 20% per year (3 cm 2 yr; 0.4 to 0.5 cm/yr diameter). Expansion rate was not affected by initial aneurysm size. During follow-up, only 3 patients (4%) required urgent operation (1 died), 26 patients (36%) died of non-aaa causes, and 26 patients (36%) underwent elective AAA repair because of progressive size increase (1 died). Elective operations were performed at the rate of 10% per year, when mean AAA size had increased to 22 cm 2 (5.1 cm in diameter). Multiple regression analysis of clinical parameters available at presentation indicated that subsequent elective AA_A repair was predicted by younger age at diagnosis and larger initial aneurysm size. As anticipated, patients who underwent surgery had more rapid aneurysm expansion (5.3 cm2/yr) compared with patients who did not undergo surgery (1.6 cm2/yr;p < 0.05). This difference was caused by more rapid expansion during later followup intervals among patients selected for operation and was not predicted by the change in aneurysm size observed during initial ultrasonographic follow-up. Final aneurysm size was predicted by initial size, duration of follow-up, and both systolic and diastolic pressure. Larger initial size, longer follow-up, and widened pulse pressure predicted larger aneurysm size. Individual A_AA expansion rates, however, were variable over time and could not be predicted by any parameters tested. Patients with small AAAs may be safely followed up if their pulse pressure is controlled and elective aneurysm repair is performed when aneurysm diameter reaches 5 to 6 cm based on sequential ultrasound size measurements. (J VAsc SURG 1990;11:260-9.) Although surgical repair is the accepted treatment of abdominal aortic aneurysms (AAAs), a careful evaluation of the risks of rupture vs operation is required in each patient. In a previous study of 67 patients with small AAAs followed up by sequential ultrasound measurements, we determined that initial aneurysm size, hypertension, and chronic obstructive pulmonary disease (COPD) were important predic- From the Section of Vascular Surgery, Department of Surgery (Drs. Cronenwett, Wall, Walsh, Zwolak, McDaniel, and Schneider and Mary L. Hawkes, RN), the Department of Radiology (Drs. Sargent and Curd), and the Department of Community and Family Medidne (Drs. Freeman and Dain), Dartmouth-Hitchcoek Medical Center. Presented at the Forty-third Annual Meeting of The Societ)~ for Vascular Surgery, New York, N.Y., June 20-21, Reprint requests: Jack L. Cronenwett, MD, Section of Vascular Surgery, Dartmouth-Hitchcock Medical Center, Hanover, NH /6/ tors of increased risk of rupture.i These results suggested that even small, 4 cm diameter AAAs could be associated with a rupture rate as high as 20% per year if hypertension or COPD was present. Despite the contribution to rupture risk of variables other than size, excellent clinical results have been reported with size alone used to determine the appropriate time for operative intervention in symptom-frcc patients. Using the criterion of 6 cm in diameter, Bernstein and Chart 2 eventually operatcd on 41% of their high-risk patients with small AAAs but reduced rupture rate to less than 5%. Other surgeons have reported comparable success using the same criterion for elective aneurysm repair, s,4 When a small AAA is detected and an initial decision is made to pursue nonopcrativc therapy, future intervention is usually influenced by increased aneurysm size. Although a number of studies have found that small AAAs expand at approximately

2 Expansion rate and outcome of aortic aneurysms cm in diameter per year, individual patients show considerable variation in aneurysm expansion rates. 1-7 This study was designed to examine the influence of various clinical parameters on aneurysm expansion and subsequent outcome in a group of patients with small AAAs initially selected for nonoperative management. Statistical techniques were applied to determine the ability to predict future aneurysm expansion and the likelihood of subsequent operation in an individual patient. METHODS All patients who underwent at least two sequential ultrasound size measurements of AAAs at the Dartmouth-Hitchcock Medical Center from 1978 to 1986 were reviewed. An AAA was defined as an enlargement of the infrarenal aortic diameter more than one and a half times the more proximal aortic diameter. Only those patients for whom a deliberate decision had been made initially to pursue nonoperative management were selected for analysis. Subsequent decisions to pursue operative therapy in these patients were made on an individual patient basis by a vascular surgeon in consultation with the primary physician. Patients who initially underwent aneurysm repair without subsequent ultrasound size measurement were excluded from further consideration. Follow-up was obtained on all patients until the time of aneurysm rupture, aneurysm operation, death, or most recent outpatient visit. Information was obtained from medical records, referring physicians, patient interviews, and death certificates. Data were coded and computerized for statistical analysis. Continuous variables were compared by Student's two-tailed t test and are expressed as the mean _ SEM. Categoric variables were compared by X 2 analysis. Pearson's r correlation and Kaplan- Meier life-table analyses were used when appropriate. Multiple regression analysis considering all possible regression models was used to develop models predictive of outcome. Ultrasonography was performed with commercially available real-time sector and linear array scanners. All patients were scanned in the supine position and in transverse and longitudinal planes. Scans were obtained with a 3.5 MHz transducer except in very thin patients, who were scanned with 5 MHz transducers. Measurements were made between the leading edges of the anterior and posterior walls (AP diameter) and between the midportions of the right and left sides of the aneurysm (lateral diameter) at its largest point. AP and lateral measurements were also obtained in the proximal abdominal aorta above the level of the aneurysm. An attempt was made to record measurements as close as possible to the same level on follow-up scans.. All original ultrasound studies were remeasured to ensure accuracy. The crosssectional area of the aorta was estimated from the formula for an ellipse: area = v (AP/2) (lateral/2). Although all patients underwent at least two ultrasound measurements separated by at least I month, most patients underwent multiple examinations at varying intervals (26%: two studies; 36%: three to four studies; 38%: five or more studies). The overall aneurysm expansion rate for each patient was calculated as the difference between the first and last ultrasound measurements divided by the elapsed time interval. Initial aneurysm expansion rate was calculated from the first two ultrasound size measurements in each patient. Final aneurysm expansion rate was calculated from the last two ultrasound size measurements in all patienl:s with at least three separate determinations. PATIENT POPULATION Seventy-three patients (54 men and 19 women) from 51 to 89 years of age (mean 70 years) with small (-<-6 cm in diameter) AAAs were initially managed nonoperatively and are analyzed in this review. Overt cardiac disease 'was present in 59% of these patients. Coronary artery disease affected 45% of patients (29% had had previous myocardial infarction, 12% had active angina, and 4% had undergone previous coronary bypass surgery), whereas 8% were treated for arrhythmias and 5% had a history of congestive heart failure. Hypertension was being treated in 45% of patients. Systolic hypertension (>140 torr) was present in 44% of patients despite treatment, whereas 20% had diastolic hypertension (>90 tort) at the time of initial evaluation. Mean blood pressure was / torr. COPD was present in 37% of patients, classified as mild to moderate in 26% and severe in 11% (forced expiratory volume in 1 second <50% predicted). Symptomatic lower extremity peripheral vascular occlusive disease (PVOD) affected 44% of patients (claudication 32%, arterial ulcer 3%, and previous bypass or amputation 10%). As judged by pedal pulses, 58% of patients had no significant PVOD, whereas 14% had reduced pedal pulses and 33% had absent pedal pulses in at least one extremity. A history of cerebral vascular disease was obtained from 20% of patients (stroke 10% and transient ischemic attacks 10%). Diabetes mellitus was present in 18% of patients (3% requiring insulin, 4% using oral agents, and 11% under dietary control). The mean serum creatinine

3 262 Cronenwett et al. Journai of VASCULAR SURGERY Table I. Aneurysm size measurements All patients NonqOerated patients Operated patients (n = 73) (n = 45) (n = 28) p < * Initial AAA size Ap diameter (cm) 4.1 _ , ,04 Lateral diameter (cm) ± NS Area (cm 2) 14 _ Ratio AAA/aorta area 3.5 _ ,4 ± NS Final AAA size AP diameter (cm) 4.7 _ ± ± Lateral diameter (cm) 4.9 _ Area (cm 2) 19 _ _ Ratio AAA/aorta area 4.7 _ _ NS Expansion rate Overall interval (mo) 34 _ AP diameter (cm/yr) 0.4 _ ± _+ 0,1 NS Lateral diameter (cm/yr) 0.5 _ ± Area (cm2/yr) 3.0 ± ± Initial ultrasound interval (mo) 10 ± 1 11 _ NS Ap diameter (cm/yr) 0.5 _ ± 0.3 NS Lateral diameter (cm/yr) 0.5 ± 0.2 0, NS Area (cm2/yr) 3.8 _ _+ 2.3 NS Final ultrasound interval (mo) AP diameter (cm/yr) 0.4 ± NS Lateral diameter (cm/yr) 0.3 ± ± NS Area (cm2/yr) 3.4 ± _+ 1, Data are mean _+ SEM. ~Two-tailed Student t test, operated vs nonoperated patients. level was 1.4 _+ 0.1 mg/dl, BUN 20 _+ 1 mg/dl, cholesterol level 228 _+ 6 mg/dl, and triglyceride level 172 _+ 30 mg/dl. Cigarette smoking was prevalent, with 59% current smokers and 18% previous smokers, whereas 23% of patients had never smoked. Of the current or previous smokers, the mean smoking history was 61 pack-years. Most AAAs were discovered initially by physical examination (53%), abdominal radiographs (27%), or abdominal ultrasonography (15%) performed for unrelated problems. Five percent of patients discovered their own aneurysm by detecting a pulsatile mass. At the time of initial aneurysm recognition, 72% of patients were symptom free whereas 23% had abdominal or back pain that was unrelated to the aneurysm but led to its detection. The rationale cited in the medical record for the initial choice of nonoperative managment in these patients was usually the small size of the aneurysm (67%). In 21% of patients the severity of cardiac disease was listed as the primary factor. Advanced age, severe COPD, malignancy, and renal failure were occasionally listed as the primary reason for nonoperative management. RESULTS Patients were followed up for a mean duration of months in this study (range 2 to 115 months; median 36 months). Mean initial AAA size was 4 cm in diameter (range 2.5 to 6 cm) and 14 cm 2 ill cross-sectional area (Table I). This was three and a half times the area of the more proximal (normal) aorta. Although aneurysms were approximately circular in cross section (mean 4.1 cm AP, 4.3 cm lateral; r = 0.91), 25% differed by more than 0.5 cm in the AP and lateral diameters and 7% differed by more than I cm. Furthermore, individual patients showed different changes in these two axes over time, For purposes of analysis, elliptic area was used to combine these bidirectional changes into a single variable. During the mean 3-year follow-up, AAA area increased 20% per year (3 cm2/yr), which corresponded to a diameter increase of 10% per year (0.4 to 0.5 cm/yr; Table I). When calculated by absolute measurements, aneurysm expansion rate was positively correlated with initial aneurysm size (r = 0.23), although this correlation was poor (p = 0.06). However, when size change was considered as a proportion of initial aneurysm size, there was no correlation between expansion rate and initial size (r = -0.06; p = 0.61). In contrast, the correlation between overall expansion rate and that observed during both the initial and final ultrasound intervals was quite high (both r = 0.67;)o < 0.001). However, there was no significant correlation between aneurysm expansion rate during the initial and

4 Volume 11 Number 2 February 1990 Expansion rate and outcome of aortic aneurysms O0 80 Overail Survival: 58%._> 60 "> 40 2O Operation-free Survival: 35% I I I I I Years I 7 8 Fig. 1. Actuarial analysis indicated a 5-year survival of 58% in patients who did not undergo aneurysm surgery and a probability of surviving 5 years without requiring aneurysrn repair of only 35%. I final ultrasound intervals (r = 0.24; no significant difference) because of the high degree of individual patient variability in aneurysm expansion rate over time. Preliminary analysis demonstrated that patients with COPD had a faster rate of aneurysm expansion (5.5 cm2/yr) than had patients without COPD (1.6 cm2/yr; p < 0.05). Patients initially selected for nonoperative management because of small aneurysm size alone subsequently demonstrated slower aneurysm expansion (1.5 vs 6.0 cm2/yr; p < 0.03), whereas patients so selected because of severe cardiac disease demonstrated more rapid aneurysm expansion (6.8 vs 2.0 cm2/yr; p < 0.04). Women had faster expansion rates than men only during the first ultrasound interval (14 vs i cm2/yr;p < 0.01). None of the other variables studied showed statistically significant differences in overall aneurysm expansion rate. Patients taking propranolol tended to have slower aneurysm expansion (0.7 vs 3.7 cmi/yr; p = 0.19), but this difference may have been the result of the small sample size. Similarly, patients with PVOD tended to have more rapid aneurysm expansion, but this could be a chance finding (5.4 vs 1.6 cm2/yr; p = 0.08). Only three aneurysms (4%) required urgent repair during follow-up. Each had been larger than 5.5 cm in diameter at last measurement. One of these aneurysms ruptured freely and led to death after surgery. The other two aneurysms were operated on urgently because of acute symptoms but had not nap- tured, and both patients survived. Twenty-six patients (36%) died ofcauscs unrelated to the aneurysm (cardiac 42%, pulmonary 27%, malignancy 12%, and other 19%). An equal number of 26 patients (36%) underwent elective surgical rcpair of the ancurysms because of increasing size during follow-up (55% aortoiliac, 36% aortoaortic, and 9% aortofemoral reconstructions). One of these patients died of hemorrhage from ~:n unexplained coagulopathy, yielding a 30-day operative mortality rate of 4% in thc elective surgical group. These operations were performed at a mean interval of 28 months after the initial AAA size measurement, when mean aneurysm size had increased to 22 cm 2 (5.1 cm in diameter). Patients who did not undergo surgery had a 5-year life-table survival of 58% (Fig. 1). Five-year elective operation rate was 4896, distributed in a linear fashion at 10% per year (Fig. 2). Operation-free survival (patients alive without operation) was only 35% after 5 years (Fig. 1). When patients who undcrwcnt surgery were compared with those who did not, patients who underwent surgery were younger (66 vs 72 years of age; p < 0.01) and less frequently current smokers (29% vs 52%;p < 0.05) and had lowcr BUN valucs (15 vs 22 mg/dl; p < 0.01). Other variables were similar in the two groups, including sex, blood pressure, previous smoking, COPD, and history of myocardial infarction and other cardiac problems. Initial aneurysm area was statistically comparable in patients who subsequently underwent surgery (16 cm 2) and

5 264 Cronenwett et al. 50 = 40 o ~ 3o o o 801 7O 60 2o 10 S- S._J ;rottrnal oz" VASCULAR SURGERY i I I I Years i t i Fig. 2. Actuarial analysis revealed a 10%/yr operation rate in patients with small aortic aneurysms initially selected for nonoperative management. those who did not (13 cm2; difference not significant). As expected, however, aneurysm area and AP and lateral diameters were significantly greater at the last ultrasound measurement in patients who underwent surgery, because this measurement influenced the decision to operate (Table I). This increased area was the result of increased lateral diameter (0.8 cm/yr in patients who underwent surgery vs 0.2 cm/yr in patients who did not; p < 0.05), whereas a change in AP diameter was statistically comparable in these groups. The overall aneurysm area expansion rate was significantly greater in patients who underwent surgery (5.3 cm2/yr) compared with patients who did not (1.6 cmi/yr; p < 0.05). Of interest was that the aneurysm expansion rate during the first ultrasound interval did not differentiate patients who subsequently underwent surgery (Table I). However, the aneurysm expansion rate during the final ultrasound interval was significantly greater in patients who underwent surgery (7.3 vs 1.1 cmg/yr; p < 0.05). A backward elimination procedure according to Cox's proportional hazards model was performed to identify clinical parameters predictive of subsequent operation, s Age, sex, blood pressure, cardiac disease, COPD, PVOD, smoking history, antihypertensive medication, and propranolol were among the variables tested in the potential model. The probability of subsequent elective aneurysm surgery based on data available at the time of initial evaluation was predicted only by age and aneurysm size. Increased age at presentation reduced the probability of subsequent operation (relative risk = 0.56 per 5 years of age; 95% confidence interval: 0.41 to 0.78; p < 0.001), whereas larger initial aneurysm size increased operation probability (relative risk = 2.0/cm diameter; 95% confidence interval: 1.3 to 3.0; p < 0.01). A combination of all possible regression models and a baclcward elimination procedure with least squares regression were used to find predictors of aneurysm expansion rate and final AAA area. No variables were found to predict expansion rate, including initial AAA area, the ratio of AAA area to the proximal aorta, blood pressure, PVOD, and COPD. Final aneurysm area, however, was predicted by initial area (p < 0.001), duration of follow-up (p < 0.001), and both diastolic (p < 0.01) and systolic blood pressure (p < 0.03). The resulting equation that expresses this predication is: final AAA area = (initial area in square centimeters) (follow-up in years) (systolic blood pressure in torr) (diastolic blood pressure in torr). This model explained 74% of the variation in final area and indicates that increased initial AAA area, increased duration of follow-up, increased systolic blood pressure, and decreased diastolic blood pressure were significantly predictive of larger final AAA area in our patients. Neither systolic nor diastolic pressure alone was statistically significant in this model, but either could be replaced by pulse pressure (systolic minus diastolic pressure) with the same predictive accuracy. Although these blood pressure variables were correlated, pulse pressure was influenced more by systolic than by diastolic pressures (systolic vs pulse pressure,

6 Volume 11 Number 2 February 1990 Expansion rate and ou~:come of aortic aneurysms [] E r5 E 0 "'o r-" f3_ X III o E v m! [] [] [] [] [] [] [] [] [] El [] Initial Diameter [] o [] 3cm 4 em [] 5cm [] [] 0 2O I I I PulsePressure(torr) ] [] i Fig. 3. Predicted time required for AAA expansion by 1 cm in diameter according to initial AAA diameter and pulse pressure, developed by regression analysis of 73 patients with small AAAs followed up with ultrasound size measurements. Representative pulse pressures were derived from the combination of systolic blood pressures of 120, 150, and 180 torr and diastolic blood pressures of 70, 80, and 90 torr. r = 0.87, p < 0.001; diastolic vs pulse pressure, r = 0.27, p < 0.02; systolic vs diastolic pressure, r = 0.71, p < 0.001). This dependence of pulse pressure on systolic pressure is explained by the larger variation in systolic pressure (mean 147 torr _+ 26 SD) than in diastolic pressure (mean 83 torr _+ 13 SD) observed in these patients. Adding the knowledge of initial AAA expansion rate (i.e., that observed between the first and second ultrasound studies) did not improve the prediction of final aneurysm area. By rearranging the regression equation, it is possible to calculate the predicted time for aneurysm expansion to any final area, based on initial AAA area and blood pressure. To demonstrate the effect of these variables graphically, the predicted time for AAA expansion by 1 cm in diameter was calculated for initial aneurysm diameters of 3, 4, and 5 cm at different pulse pressures derived from representative systolic pressures of 120, 150, and 180 torr and diastolic pressures of 70, 80, and 90 torr (Fig. 3). These results indicate that the time for AAA diameter expansion by 1 cm varies from 0 to 11 years, with higher pulse pressure associated with significandy more rapid expansion. DISCUSSION Our current study population of 73 patients with small aortic aneurysms from New England is remarkably similar to our previous study population of 67 patients from Michigan reported in Age, sex, aneurysm size, expansion rate, and follow-up were nearly identical in these groups. However, in our earlier report only 3% of patients underwent elective aneurysm repair compared with 36% in the current study, despite comparable increases in aneurysm size. This difference appears to explain the higher (I8%) incidence of aneurysm rapture or acute expansion observed in our earlier study, compared with only 4% in our present report (Table II). This apparent difference in patient management strategy reflects more direct involvement by vascular surgeons in our current study and more exclusive management by primary care physicians in our earlier, more "observational" report. Although rigorous criteria for operation were not applied during the period of this retrospective review, the data indicate that most initially small (3 to 4 cm in diameter) AAAs were operated on when they achieved a dj[ameter of 5 cm, except in very high-risk patients. Three previous reports of patients with small aneurysms managed by similar criteria have demonstrated remarkably similar results (Table II).2-4 A summary of these results yields a total of 378 patients with small AAAs initially treated nonoperatively and followed lap for a mean of 31 months. During this follow-up, elective operation was performed in 38% of patients, whereas only 6% had acute expansion or aneurysm rupture. Death from other causes occurred in 29% of patients, whereas 27% were alive with their aneurysm at follow-up,

7 266 Cronenwet~ et al. journal of VASCULAR SURGERY Table II. Outcome of patients with small AAAs initially treated nonoperativcly Aneurysm Mean Alive with rupture~acute Mean age follow-up aneurysm Died of other Elective operation expansion Author n (yr) (mo) (%) causes (%) (% ~mortality) ~ (%/mortality) Cronenwett et a /0 18/67 Bernstein and Chart /5 4/67 Littooy et al.a /3 6/100 Sterpetti et al /9 6/67 Present study /4 4/33 Meanll /5 5/67 ~Thirty-day mortality rate after elective aneurysm operation.?thirty-day mortality rate after aneurysm rupture (whether operated or not) plus emergent procedures Ibr acute aneurysm expansion. ::Observational study with only 3% subsequent intervention. Studies with planned elective aneurysm repair when 5 to 6 cm diameter is reached. HCombined results of planned intervention of studies with footnote. 'Yl'otal. resulting in an overall mean 5-year survival rate of 54%. These combined results provide an accurate estimate of the anticipated outcome of contemporary management of patients with small AAAs who arc initially treated nonoperatively and carefully followed up with sequential ultrasound studies until a diameter of 5 to 6 cm is reached. Since the 5% elective operative mortality rate in these combined reports compares favorably with the 6% per year rupture rate reported in our previous observational study, it does not appear that these criteria are too liberal. The fact that only one ancurysm ruptured in our present study strongly suggests that increased application of elective ancurysm repair reduced rupture rate. However, this prevented us from identifying clinical variables that might be predictive ofancurysm rupture. Thus we were not able to test our previous finding that increased inkial ancurysm size, hypertension, and COPD are three important predictors of rupture) In our present study, increased pulse pressure was associated with increased aneurysm expansion, but COPD was not included in the predictive model, although AAAs in patients with COPD did expand more rapidly by univariate analysis. We previously speculated that COPD might be associated with AAA rupture through the joint mechanism of increased protease activity in these patients.1 Dobrinet al.9 have shown that aneurysm expansion is associated with decreased elastin content, whereas aneurysm rupture is associated with decreased collagen content. Although there are conflicting data concerning the precise role of elastin and collagen in aneurysm rupture and expansion, ~ a differential effect of these proteases might explain why COPD was predictive of AAA rupture but not AAA expansion in our studies. Despite the powerful statistical techniques used in this analysis, we could not identify any clinical parameter that could predict specific aneurysm expansion rate. We were unable to verify the previous report of Sterpetti et al.4 that PVOD and the ratio of initial aneurysm size to aortic diameter were predictive of aneurysm expansion rate. In fact we found that AAA expansion rate did not correlate with initial aneurysm size. Proportional size change (i.e., 20% area increase or 10% diameter increase per year) reflects AAA expansion more accurately than it does absolute change (i.e., 0.5 cm in diameter/yr) because the latter is somewhat influenced by initial size. Despite these negative results, we were able to predict final aneurysm area quite accurately, with variables that are available at the time of initial diagnosis. Not surprisingly, we found that larger initial aneurysm size and longer duration of follow-up were important predictors of final aneuvsm size. In addition, we identified blood pressure as a significant predictor, such that either increased systolic pressure or decreased diastolic pressure was associated with increased aneurysm expansion. Although the direction of the effect of diastolic blood pressure initially appears counterintuitive, this effect can be explained by the greater importance of a widened pulse pressure; that is, at any fixed systolic pressure, decreasing diastolic pressure causes increased pulse pressure. This result suggests that the magnitude of cyclic stress imposed on the aortic wall during each heartbeat, a function of pulse pressure, is more central to AAA expansion than is the baseline diastolic pressure. We believe that this is the first demonstration that systolic hypertension with a resultant wide pulse pressure is an important risk factor for AAA expansion,, similar to the risk of systolic hypertension in the de-

8 Volume 11 Number 2 February 1990 Expansion rate and outcome of aortic aneurysms 267 velopment of generalized complications of atherosclerosis, n To develop appropriate operative criteria for patients with small AAAs, it is important to note that neither final AAA size nor eventual operation was predicted by the initial aneurysm expansion rate as calculated from the first to second ultrasound measurements. This observation, plus the large variation of individual aneurysm expansion rates over time, emphasizes the importance of prolonged, sequential ultrasound follow-up. Based on these results, we recommend the use of final AAA diameter, rather than expansion rates between sequential size measurements, as the appropriate criterion for operative intervention. Aneurysms eventually selected for operation in this study were significantly larger than those in patients who did not undergo surgery, as noted previously by other investigators in similar populations? However, this difference was the result of 'more rapid aneurysm expansion during later ultrasound follow-up intervals, when the expansion rate of operated aneurysms (7.a cm2/yr) was significantly greater than that of nonoperated aneurysms (1.1 cmi/yr; p < 0.05). Serial ultrasound measurements of aneurysm diameter provide a clinically reliable method to follow up patients with small AAAs? 2 Although computed tomography is an acceptable alternativc, it is more expensive and requires radiation exposure, s We have not had difficulty obtaining accurate, reproducible measurements with ultrasonography. Although AP diameter is obtained more accurately by ultrasonography bccause of the insonating angle of thc ultrasound beam, lateral diameter may be more important? a We observed that elliptic aneurysms had their long axis in the lateral dimension more frequently and previously noted that smaller aneurysms (3 to 4 cm in diameter) initially expanded more rapidly in the lateral direction? Because rupture is also more frequent in the lateral direction, 1 we believe that both dimensions should be followed. We were unable to confirm the previous suggestion by Leach et al.6 that propranolol treatment is associated with less rapid aneurysm expansion. In a study of 27 patients with small AAAs, these authors found less rapid aneurysm expansion in patients receiving [3-blockade (independent of blood pressure), which was nearly significant (p = 0.08). They speculated that this effect might be related to increased aortic tensile strength as a result of the ability of propranolol to stimulate lysyl oxidase activity, which promotes collagen and elastin cross-linking. 6 Further studies will be required to substantiate this hypothesis, particularly because of the known ability of [3- blockade to reduce pulse pressure and dp/dt, a potentially confounding effect. Rupture of AAAs remains a significant problem despite improved early detection by ultrasonography. Furthermore, the overall incidence of these aneurysms appears to be inc;easing independent of increased detection. 14 The ultimate goal of management in patients with small AAAs is to select for nonoperative therapy those patients who will die of other causes before aneurysm rupture and to operate on the remaining group. Aneurysm size has been the usual discriminator for this approach, with AAAs smaller than 5 or 6 cm in diameter usually being followed up by serial ultrasonography. Our present study strongly supports this approach because only 1% of 73 patients died of aneurysm rupture when elective operation was performed at the rate of 10% per year, with an elective: operative mortality rate of only 4%. With this approach, AAA rupture or elective surgery contributed only 3% to the S-year mortality rate observed in these patients. Based on the results of our present study, we recommend sequential (6-month interval) ultrasound measurement of AAAs less than 5 cm in diameter (or 6 cm in diameter in some high-risk patients). Systolic hypertension should be treated aggressively to minimize pulse pressure and the resultant aortic stress. By using our regression model (Fig. 3), it is possible to predict the expected time for AAA expansion by i cm in diameter for any given pulse pressure. If this result is compared with the patient's life expectancy based on other underlying risk factors, it is possible to make an informed judgment concerning the likelihood that aneurysm expansion to 5 (or 6) cm will occur during the patient's lifetime. If this is likely, elective surgical repair of the AAA before it reaches this absolute size criterion may be beneficial, particuarly in younger patients with hypertension, because surgical treatment of smaller aneurysms is probably less complicated, is Finally, we believe that COPD is an independent predictor of AAA rupture, although not of gradual expansion. Until the importance of this factor is clarified further, we adopt a more aggressive elective operative approach toward patients with COPD. Ideally, a prospective, multicenter study will be conducted, not only to determine accurately clinical risk factors that affect AAA rupture but also to compare aggressive medical management of hypertension with early surgical intervention. Until then the guidelines sug-

9 268 Cronenwett et al. Journal of VASCULAR SURGERY gested above have yielded satisfactory results in our recent experience with these patients. REFERENCES 1. Cronenwett JL, Murphy TF, Zelenock GB, et al. Actuarial analysis of variables associated with rupture of small abdominal aortic aneurysms. Surgery 1985;98: Bemstein EF, Chan EL. Abdominal aortic aneurysm in highrisk patients: outcome of selective management based on size and expansion rate. Ann Surg 1984;200: Littooy FN, Steffan G, Greisler HP, White TL, Baker WH. Use of sequential B-mode ultrasonography to manage abdominal aortic aneurysms. Arch Surg 1989;124: Sterpetti AV, Schultz RD, Feldhaus RJ,Cheng SE, Peetz DJ Jr. Factors influencing enlargement rate of small abdominal aortic aneurysms. J Surg Res 1987;43: Delin A, Ohlsen H, Swedenborg J. Growth rate of abdominal aortic aneurysms as measured by computed tomography. Br )r Surg 1985;72: Leach SD, Toole AL, Stem H, DeNatale RW, Tilson MD. Effect of [3-adrenergic blockade on the growth rate of abdominal aortic aneurysms. Arch Surg 1988;123: Taylor LM, Porter IM. Basic data related to clinical decision making in abdominal aortic aneurysms. Ann Vase Surg 1986;1: Cox DR. Regression models and life tables. J R Stat Soc [Set B] 1972;34: Dobrin PB, Baker WH, Gley WC. Elastolytic and collagenolytic studies of arteries. Arch Surg 1984; 119: I0. Cohen JR, Mandell C, Margolis I, Chang J, Wise L. Altered aortic protease and antiprotease activity in patients with ruptured abdominal aortic aneuusms. Surg Gynecol Obstet 1987;164: Fergnson Jl III, Otelio SR. Systolic, diastolic, and combined hypertension: differences between groups. Arch Intern Med 1986; 146: Graeve AH, Carpenter CM, Wicks JD, Edwards WS. Discordance in the sizing of abdominal aortic aneurysm and its significance. Am J Surg 1982;144: McGregor JC. Abdominal aortic aneurysms: ultrasonic measurement of the transverse diameter and its prognostic significance in the light of pathological observations. Postgrad Med J 1977;53: Bickerstaff LK, Hollier LH, Van Peenen HJ, Melton LJ III, Pairolero PC, Cherry KJ. Abdominal aortic aneurysms: the changing natural history. J VASe SUV, G 1984;1: McCabe CJ, Coleman WS, Brewster DC. The advantage of early operation for abdominal aortic aneurysm. Arch Surg 1981;ii6: i DISCUSSION Dr. Eugene F. Bernstein (LaJolla, Calif.). This study has provided new data regarding the selection of patients for elective surgery that specifically focuses on the issue of the good-risk patient with a small aneurysm. To put this into perspective, we should see how it impacts the major unanswered questions about aneurysm pathogenesis and the indications for surgery. Our overall goal is to reduce the number of deaths caused by aneurysm rupture. To do this we would like to identify aneurysms as early as possible and operate on patients when they are in the best condition and therefore have the lowest operative mortality rates. On the other hand, if we could predict which patients with small aneurysms would remain stable for many years or decades, we might spare those patients the risks and discomfort of elective surgery. Thus we would really like to know why aneurysm expansion starts, know whether there are indexes in the earliest studies we can do that predict future expansion or rupture, and then create a logical management plan. Individualizaiton of management is critical in balancing those factors. Clearly the prospect of routine population screening appears attractive in identifying the lesion early and might well be cost-effective if performed by limited computed tomographic (CT) or ultrasound methods. However, without a dear indication of the risk of individual aneurysm expansion potential, justification for routine elective surgery for the 3 to 4 or perhaps even 5 cm aneurysm continues to be debatable. My practice has become more aggressive in the past two decades and I now operate dec- tively on 5 cm aneurysms in good-risk patients, 6 cm aneurysms in poor-risk patients, and selectively on aneurysms between 4 and 5 cm. This report provides us with confirmatory data regarding the mean aneurysm expansion rate, which is between 4 and 5 mm in diameter per year, the wide variability in that rate, and some additional risk factors, tt emphasizes cross-sectional area, rather than diameter, which has been used in most prior reports. My first question relates to this area index, because it involves multiplying two dimensions, each of which is measured by ultrasound and variables within a few millimeters. Is there evidence that a calculated area is a better index than the largest transverse diameter? Is it as reproducible in measuring? Second, you emphasize the ratio of the aneurysm crosssectional area to the proximal normal aorta. Although this ratio has been discussed in the past, I believe your report represents the first in which the ratio of aneurysm to aortic area has been used to describe a group of patients who are followed up nonoperatively with this index. Are there data regarding the likelihood of rupture with this ratio? Finally, this material differs from that of your prior report because more patients were managed by vascular surgeons and therefore 36% were operated on for the enlargement or development of symptoms. Nevertheless, the overall 5-year survival rate was only 58%, just 6% higher than that of your Michigan experience, in which elective surgery was performed in only 2 of 67 cases. Many other recent series show a considerably higher 5-year survival rate, ranging up to 80% for aneu-

10 Volume 11 Number 2 February 1990 Expansion rate and outcome of aortic aneurysms 269 rysms routinely operated on. Was there a difference in the patient population between the Michigan and Dartmouth groups? Do you believe you would be even more aggressive now regarding workups in these patients for both cardiac and aneurysmal lesions? Finally, I must comment that this was not a prospective trial but a retrospective study that used ultrasonography, which is a less precise and less objective modality than CT or magnetic resonance imaging in providing cross-sectional area information and does not provide accurate thrombuslumen information. Regarding the latter factor, at our institution Drs. Krupski and Bass have recently reviewed serial CT scans in 30 patients who did not undergo surgery for at least 6 months in an effort to identify predictive indexes for expansion. Four indexes were associated with expansion, although only two achieved statistical significance in this preliminary study: thrombus/lumen ratio and thrombus/aneurysm ratio. Do you have any experience looking at thrombus/aneurysm area or thrombus/lumen or aneurysm ratios as a means of identifying which aneurysms are likely to expand in the near future? Finally, what is your current practice in dealing with the good-risk patient with a 4.5 cm aneurysm? Dr. Cronenwett. Although it is possible that crosssectional area might introduce more variance, the primary reason for using this approach was to combine the twodiameter indexes into a single variable for statistical analysis. By using regression techniques, we found that area was actually a more accurate predictor of final aneurysm size than either the anteroposterior or lateral diameter measurements alone. We did not find that the ratio of aneurysm to aortic diameter was more predictive of change of final size than was initial aneurysrn size alone. It is possible that this resuk may be caused by the introduction of more variance by combining these measurements, as you suggested could be possible for area. Because we attempted to analyze and predict the outcome of patients with small aneurysms in this study, we considered operation as an end point and we did not compare the subsequent survival rates after operation with patients who did not undergo surgery. Comparison with our previous study, however, suggests that aneurysm rupture was reduced significantly by elective intervention. To address your question, a prospective study would be required to compare nonoperative and operative strategies in patients with small aneurysms. Although this was not a prospective study, I believe that the combined results of the four recent studies reviewed in our article are representative of the low mortality and reduced rupture rates possible with current techniques and elective procedures. We have previously looked at thrombus area within the aneurysm and have not found it to be predictive of rupture when analyzed by ultrasonography. I would indicate that it is important to analyze these factors with multivariate techniques that avoid identifying individual factors that are not themselves significant but only correlated with more important predictors. Finally, we agree in general with your management plan: that is, we operate on most aneurysms when they reach the size of 5 cm in diameter. We will occasionally operate on smaller aneurysms, especially in patients with systolic hypertension that is difficuk to control or young, good-risk patients whose aneurysm will likely expand to a significant size during their active lifktime. Dr. Colleen M. Broplhy (New Haven, Conn.). Dr. Cronenwett, you have determined that long-term ultrasound follow-up combined with judicious operative intervention achieved a low mortality rate in patients with small aneurysms. The dissatisfying aspect of this approach is that it entails a watch and wait: approach without the benefit of therapeutic intervention. Work with two animal aneurysm models, the lathyritic turkey and the blotchy mouse, has suggested that propranolol delays aneurysm formation. In addition, experiments in both animal aneurysm models suggest that the propranolol-mediated delay in aneurysm expansion may be caused by a direct effect oft matrix metabolism rather than the known hemodynamic effects of the drug. In addition to our work with the blotchy mouse, we also have data based on a retrospective analysis of patients followed up nonoperatively that suggests that [3-blockers delay aneurysm expansion. Your study on the variables that affect expansion rate and outcome of small aneurysms as reported in the abstract, showed that patients taking propranolol tended to have less rapid aneurysm expansion, namely 0.7 cm ~ per year in the propranolol group vs 3.7 cm 2 per year in patients not receiving [3-blockers. In spite of this sixfold difference, your results failed to achieve statistical significance. Could you elaborate on your data? Was this secondary to an outlier? Also, do you believe that there are enough data to initiate a clinical trial of propranolol on small aneurysms that are being followed up with serial ultrasound examinations? I voice this question with the concern raised by Dr. Hollier that nonoperative treatment of small aneurysms with or:without pharmacologic intervention may delay treatment of aneurysms until the patients are older and poorer candidates for definitive surgical repair. Dr. Cr0nenwett, To address your comments specifically, we analyzed 16 patients taking propranolol. You are correct that by univariate analysis there appeared to be a trend among patients taking propranolol to have less rapid aneurysm expansion. However, when analyzed with multiple regression techniques, this was not significant. Although this did not appear to be the result of "outliers," the number of patients involved is too small to be confident of this conclusion. In designing a human study, it would be important to realize the potential complicating effects of propranolol on the important factors of both systolic blood pressure and dp/dt, in addition to potential effects on collagen crosslinking. Your studies provide more basis for a clinical trial of propranolol in these patients than do ours.