View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Eur J Vasc Endovasc Surg 18, 96–104 (1999) Article No. ejvs.1998.0786

Morphology of Ruptured Abdominal Aortic Aneurysms

J. Golledge∗, J. Abrokwah, K. N. Shenoy and R. H. Armour

Department of Surgery, The Lister Hospital, Corey’s Mill Lane, Stevenage SG1 4AB, U.K.

Objectives: it has been suspected that the position at which aneurysm rupture occurs would affect outcome. The aim of this prospective study was to relate the morphology of the ruptured abdominal aortic aneurysm (RAAA) to the clinical characteristics and outcome of the patient. Design, patients and methods: over a five-year period 46 patients with RAAAs (11 identified at operation and 35 at post-mortem (PM)) were examined to identify the number and site of aneurysm rupture, the characteristics of the aneurysm (site, shape, length, diameter and associated iliac aneurysm) and the presence and site of retroperitoneal or intraperitoneal haematoma. The patients were also assessed for a range of associated medical conditions. Univariate analysis was used to identify variables predicting the site of aneurysm rupture and survival following rupture. Results: most aneurysm ruptures (73%) occurred in the middle third where the aneurysm diameter was greatest. Relatively few ruptures (13%) were in the left posterior quadrant of the aneurysm. A history of chronic lung disease (p=0.04) significantly predicted the quadrant of aneurysm rupture, while a history of carcinoma (p=0.02) predicted the level of rupture. Nine of the 46 (20%) patients survived. Survival was predicted by the transverse site of rupture (p=0.0004) and the absence of ischaemic heart disease (p=0.02). Conclusion: these data suggest that the site of aneurysm rupture, which may be in part predicted by the clinical characteristics of the patient, influences outcome.

Key Words: Aortic aneurysm; Rupture; Site.

Introduction and therefore survival. In this study we have assessed the importance of the rupture site of abdominal aortic At present, there is disagreement about which patients aneurysms. should be selected for elective abdominal aortic an- eurysm repair, and the results of the small aneurysm trial are awaited with interest.1,2 Further understanding of the mechanisms leading to rupture of abdominal Methods aortic aneurysms would enable better selection of patients for endovascular or open repair and identify Over a five-year period, patients were recruited from possible therapeutic targets for the medical control of two sources: Firstly ruptured abdominal aortic an- aneurysm rupture. Aneurysm rupture occurs at the eurysms (RAA) undergoing surgical intervention site at which the stress exerted on the wall from within (n=11), and secondly all similar cases identified at overcomes the ability of the wall to resist these forces. post-mortem (PM) examination (n=35). Clearly there Aortic wall stress is dependent on a complex relation- would have been other patients who died suddenly ship between aortic pressure and wall thickness.3,4 in the area, where the diagnosis of abdominal aortic While the aortic wall strength is determined by the aneurysm was not suspected and who did not undergo elastin and collagen content.5–8 Thus the site of an- post-mortem examination. Of the 35 post-mortem eurysm rupture could give important clues on the cases, 12 patients died suddenly at home, eight had process of rupture. In addition, rupture site might clear preceding symptoms but failed to seek medical determine the likelihood of free intraperitoneal rupture help, four reached hospital but failed to be resuscitated, four patients were felt unsuitable for operation, three ∗ Please address all correspondence to: J. Golledge, 35 Sullivans died during transit (two from home to The Lister and Reach, Walton on Thames, Surrey, London KT12 2QB, U.K. one during transfer to another hospital), three were

1078–5884/99/080096+09 $12.00/0  1999 W.B. Saunders Company Ltd. Morphology of Ruptured Abdominal Aortic Aneurysms 97

admitted to hospital and died without the correct Table 1. Site of rupture along abdominal aortic aneurysm. diagnosis being made and one patient died in the operating theatre (following anaesthetic induction Site of rupture Operative cases PM cases Total prior to laparotomy). For these patients the median Upper third 4 3 7 time from symptoms to death was 2 hours, range Middle third 7 28 35 Lower third 0 6 6 0–330 hours. Eleven cases were excluded for the following reasons: inability to identify the site of rupture at operation (n=6, three survived and were discharged home and three died within 30 days), transfer to Results another hospital (n=3), ruptured thoracoabdominal aneurysm (n=1) and ruptured non-aneurysmal aorta The median age of the patients was 73.5 (range 60–93) (n=1). Local ethical approval was obtained. Operative years with the majority being male (40 of 46). The and post-mortem specimens were examined to identify incidence of associated disease was as follows: the following: the site of the aneurysm rupture in ischaemic heart disease (31%), hypertension (37%), the transverse (recorded as 1 to 12 o’clock and later cerebrovascular disease (13%), chronic obstructive air- categorised as right or left anterior or posterior quad- way disease (13%), renal failure (11%), carcinoma (9%; rant) and longitudinal (upper, middle or lower third) bronchial carcinoma 2, colonic carcinoma 1, jejunal dimensions; the number and length of ruptures; the carcinoma 1), diabetes mellitus (7%) and liver cirrhosis presence and principal position (right, central, left or (2%). pelvic) of retroperitoneal haematoma; the presence and volume of intraperitoneal haematoma; the pres- ence of aortocaval or aortogastrointestinal fistulae; the shape (fusiform, saccular or hourglass), site Characteristics of abdominal aortic aneurysm and rupture (infrarenal, juxtarenal or suprarenal), length and diameter of abdominal aortic aneurysm (by calliper Most aneurysms were infrarenal (40, 87%), four were measurement); the presence and characteristics of juxtarenal (9%), and two were suprarenal (4%). Ninety- associated iliac aneurysms. The following factors six per cent of aneurysms (44) were fusiform, one was saccular and one hourglass in shape. All aneurysms were also identified by a combination of history, measured 5 cm or more in transverse diameter, the clinical examination, blood tests and examinations median diameter was 9 cm (range 5–14 cm). The me- performed at PM: age, gender, ischaemic heart dian aneurysm length was 11 cm, range 7.5–18 cm. In disease (history, ECG or PM evidence of myocardial two cases (one identified at operation and one at PM) infarction), hypertension (systolic BP>180 mmHg, there were two sites of aneurysm rupture; in these diastolic BP>100 mmHg or ECG/PM evidence of patients, the principal site of rupture was taken as the left ventricular hypertrophy), renal failure (serum larger. In all other cases (96%) there was one site creatine>120), diabetes mellitus, cerebrovascular of rupture. The position of these 48 ruptures in the disease (previous stroke or TIA), previous or recent longitudinal and transverse dimensions are shown in carcinoma, liver cirrhosis, chronic lung disease (em- Table 1 and Fig. 1. Most ruptures were in the middle physema, chronic bronchitis or bronchiectasis; diag- third of the aneurysm (73%), while in the transverse nosed by history or PM examination of lungs), time dimension ruptures were distributed all around the from symptoms to death. Outcome was defined as circumference of the aneurysm, with relatively few survival for those patients undergoing successful op- ruptures in the left posterior quadrant (Fig. 1). Sur- erations and being discharged home well, or death for prisingly, most ruptures identified at operation (46%) those patients having unsuccessful operation or whose were in the left posterior quadrant. aneurysms were identified only at PM. Data was stored prospectively on record cards and later transferred to a computer spreadsheet (Microsoft Excell, Microsoft Limited, Reading, U.K.) and statistical Retroperitoneal and intraperitoneal haematoma and program (Statview, Abacus Concepts, Berkeley, CA, relation to rupture site U.S.A.). Univariate (Chi-squared with Yates correction, Mann–Whitney U-test and Kruskal–Wallis test) ana- All except one of the patients had a huge retro- lysis was used to identify variables predicting the site peritoneal haematoma while only 19 (41%) had intra- of aneurysm rupture and survival. peritoneal blood (in nine of these 19 cases this was

Eur J Vasc Endovasc Surg Vol 18, August 1999 98 J. Golledge et al.

Fig. 1. The distribution of rupture sites in the transverse dimension for 46 abdominal aortic aneurysms. In two cases there were two sites of rupture. Shown are numbers for PM (first number) and operative cases.

blood-stained fluid rather than frank blood). The one transverse dimension and the position of the retro- patient with no retroperitoneal or intraperitoneal peritoneal haematoma, p=0.0001 (Table 2). haematoma had an aortocaval fistula. Intraperitoneal bleeding was identified in seven of 11 patients (64%) undergoing operation compared to 12 of 35 (34%) aneurysms identified at PM, p=0.04. Table 2 shows Factors associated with the site of aneurysm rupture the principal site of the retroperitoneal haematoma as (Table 3) right abdominal (54%), central (7%), left abdominal (35%) or pelvic (2%). There was a very close re- Table 3 presents the clinical characteristics and the lationship between the site of aneurysm rupture in the aneurysm morphology for patients according to the

Eur J Vasc Endovasc Surg Vol 18, August 1999 Morphology of Ruptured Abdominal Aortic Aneurysms 99

Table 2. Relationship between site of rupture and retroperitoneal haematoma.

Position of retroperitoneal haematoma Right Left Central Pelvic None

Site of rupture Right anterior 11 4 0 1 0 Right posterior 12 0 1 0 1 Left anterior 2 8 2 0 0 Left posterior 0 4 0 0 0 Total 25 16 3 1 1

Site of rupture predicted position of retroperitoneal haematoma, p=0.0001 by Chi-squared.

Table 3. Variables predicting site of abdominal aortic aneurysm rupture.

Left anterior Left posterior Right posterior Right anterior Variable (n=12) (n=4) (n=14) (n=16) p value

Median age (range, years) 76.5 (60–93) 71.5 (67–80) 72.5 (66–85) 74 (61–88) NS Male 10 (83%) 4 (100%) 13 (93%) 13 (82%) NS Ischaemic heart disease 3 (25%) 0 4 (29%) 7 (44%) NS Diabetes 0 1 (25%) 1 (7%) 1 (6%) NS Renal failure 2 (17%) 0 0 3 (19%) NS Cerebrovascular disease 3 (25%) 0 3 (21%) 0 NS COAD 0 0 4 (29%) 2 (13%) 0.04 Hypertension 5 (42%) 1 (25%) 6 (43%) 5 (31%) NS Carcinoma 0 1 (25%) 1 (7%) 2 (13%) NS Iliac aneurysm 7 (58%) 0 6 (43%) 4 (25%) NS Median aneurysm diameter (range, cm) 8.5 (6–13) 9.5 (7–10) 8.5 (5–10) 8.8 (6–14) NS Median aneurysm length (range, cm) 11 (8–18) 12 (8–15) 10 (7.5–17) 11 (8–14) NS Survival 2 (17%) 4 (100%) 1 (7%) 2 (13%) 0.0004

quadrant in which their aneurysm ruptured. The only successful cases was 4 years (range 3 to 9.8 years). The variable which predicted the site of aneurysm rupture eventual cause of death was available for six of these was the presence of chronic obstructive airway disease patients: bronchopneumonia 2, myocardial infarction (COAD). All the six patients with chronic lung disease 2, ruptured iliac aneurysm 1. The most significant suffered ruptures on the right side in an arch from 7 factor predicting outcome was the transverse site of to 10 o’clock: three occurred at the 7 o’clock position, aneurysm rupture (Table 4). All four patients in whom while the other three were at 8, 9 and 10 o’clock the principal rupture site was in the left posterior respectively, p=0.04. A similar analysis was performed quadrant survived compared to only one of the 14 to identify variables predicting the longitudinal site patients who had a right posterior rupture, p=0.0004 of the rupture, which was the middle third in 73% of (Fig. 2). Six of the nine patients surviving had left- patients. The only significant predictor was the recent sided ruptures compared to only 10 of the 37 patients treatment of or unexpected finding at PM of carcinoma, who died. Ischaemic heart disease and associated iliac = p 0.02. Of the four patients with carcinoma, three of aneurysm predicted a less favourable outcome (Table the ruptures occurred outside the middle third of the 4). Despite the wide range of times between the onset aneurysm (1 upper and 2 lower third). of symptoms to death, univariate analysis identified 2 factors predicting a short time between the onset of symptoms and death: one rather than two sites of Outcome following aneurysm rupture and variables rupture (p=0.01) and the absence of cerebrovascular predicting outcome disease (p=0.03). The median time from symptoms to death was 56 (range 0–312) hours for patients with Nine of the 46 patients survived rupture of their cerebrovascular disease compared to 2 (range 0–330) aneurysm. The median length of survival for the nine hours for patients without cerebrovascular disease

Eur J Vasc Endovasc Surg Vol 18, August 1999 100 J. Golledge et al.

Table 4. Variables predicting survival from ruptured abdominal aortic aneurysm.

Variable Survived (n=9) Died (n=37) p value

Median age (range, years) 72 (63–88) 74 (60–93) NS Male 9 (100%) 31 (84%) NS Ischaemic heart disease 0 14 (38%) 0.02 Diabetes 1 (11%) 2 (5%) NS Renal failure 1 (11%) 4 (11%) NS Cerebrovascular disease 0 6 (16%) NS COAD 1 (11%) 5 (14%) NS Hypertension 2 (22%) 15 (41%) NS Carcinoma 1 (11%) 3 (8%) NS Iliac aneurysm 1 (11%) 16 (43%) 0.07 Supra- or juxtarenal aneurysm 0 6 (16%) NS Median aneurysm diameter (range, cm) 9 (6–10) 8.5 (5–14) NS Median aneurysm length (range, cm) 13 (11–15) 10.3 (7.5–18) NS Median length of rupture (range, cm) 4 (1–5) 2.5 (1–9) NS Left-sided rupture 6 (67%) 10 (27%) 0.0004 Right-sided rupture 3 (33%) 27 (73%) Upper third rupture 3 (33%) 3 (8%) 0.09 Middle third rupture 6 (67%) 29 (78%) Lower third rupture 0 5 (14%) Intraperitoneal blood 7 (78%) 12 (32%) 0.01

(p=0.03). The two patients with two sites of rupture middle third of the aneurysm is in agreement with died 60 and 330 hours following onset of symptoms. the study of Darling and colleagues11 (Fig. 3). These Interestingly, the second and minor site of rupture data suggest the importance of wall stress in de- in these two cases were both in the left posterior termining the site of rupture. Computer models of quadrant. aneurysms have been used to calculate wall stresses using the method of finite element analaysis.4,15 Using this technique it has been demonstrated that wall Discussion stress is maximal at the point of maximum aneurysm diameter, i.e. the midpoint, although this may not 4,15 Autopsy and clinical series have identified the dia- always be the case for circumferential stresses. An- meter of an abdominal aortic aneurysm as an im- eurysm shape is another factor which has been dem- 16 portant risk factor for rupture.9–12 For example, the onstrated to determine wall stress. Classification of annual rupture rate for an aneurysm of 10-cm diameter aneurysm shape is subjective; in our study we dif- has been estimated to be as high as 90%, compared ferentiate three broad categories, i.e., fusiform, saccular to 0 to 5% for aneurysms less than 5 cm in diameter.13 and hourglass; however, more subtle types may be Not surprisingly, therefore, the selection of patients discernible. Only two patients had aneurysms which for elective operation has been largely based on were not fusiform, one saccular and one hourglass. aneurysm diameter in relation to operative risk.14 The site of rupture for these two non-fusiform an- However, it is becoming clear that factors other than eurysms was unusual in being outside the middle diameter alone are important in determining when third; one in the upper and one in the lower third. rupture will occur.13 A better understanding of Hypertension is another factor which would be ex- rupture risk may be gained by a consideration of pected to influence wall stress and has been related the variables influencing wall stress and wall tensile to the risk of aneurysm rupture.9,10 Varying definitions strength, since at the point when wall stress over- are used for hypertension and any single reading of comes the tensile strength of the aortic tissue rupture blood pressure may be inaccurate, particulary if recent will occur. readings prior to aneurysm rupture are unavailable. In this study we have assessed clinical and ana- We therefore looked for evidence of long-term hyper- tomical variables in patients with RAAA in order to tension, such as ECG or PM evidence of left ventricular gain clues as to the events taking place at the point of hypertrophy, in defining this condition. This definition rupture. The finding that most ruptures occur in the of hypertension different to that used in other studies

Eur J Vasc Endovasc Surg Vol 18, August 1999 Morphology of Ruptured Abdominal Aortic Aneurysms 101

Survived 2 Survived 2 death 14 death 10 Mt 2.5 (0–330) h Mt 3.5 (0–312) h

16 (35%) 12 (26%)

14 (30%) 4 (9%)

Survived 1 Survived 4 death 13 death 0 Mt 2.0 (0–72) h

Fig. 2. The outcome in relation to the site of rupture of abdominal aortic aneurysm. Mt is the median time from commencement of symptoms to death, with range in brackets. Note the safety of left posterior quadrant ruptures and relatively rapid death following right posterior rupture.

may explain the lack of correlation between this factor Other investigators have demonstrated an increased and the site of aneurysm rupture found in our study risk of aneurysm rupture in patients with chronic lung (Table 3). disease.9,10 Using Cox proportional hazard analysis Increased wall stress alone does not explain all Cronenwett and colleagues estimated a five-year rup- the characteristics of the ruptured abdominal aortic ture risk of 34% for patients with small aneurysms aneurysm. For example, as the aorta dilates the thick- who had severe COAD, compared to only 2% in those ness of the wall actually increases, a change likely to patients without lung disease.9 It was suggested that reduce wall stress.17 Remodelling of the aorta leads to this association between lung disease and risk of a thicker but weaker wall characterised by a collagen- aneurysm rupture may be due to increased proteinase laden fibrotic adventitia.17 Our analysis of factors activity in the aneurysms of this group of patients.9,10 associated with the rupture site indicates the import- Studies of human aortic biopsies have suggested a ance of patients’ associated medical conditions in role for matrix metalloproteinase 2 (MMP-2, gel- predicting the site of the rupture. In all, six patients atinase A) in the expansion of small aortic aneurysms, with chronic lung disease rupture of the aneurysm while MMP-9 (gelatinase B) may be more important took place on the right side, while for three of the in the final stages of aortic wall weakening prior four patients with an incidentally detected or past to rupture.17 Alterations in the activity of matrix history of carcinoma rupture was outside the middle metalloproteinases and their tissue inhibitors (TIMPS) third of the aneurysm. While there are many possible have been demonstrated in patients with emphysema explanations for this association it is tempting to and in association with a variety of different car- suggest that these comorbid conditions are reflected cinomas.18,19 Subtle changes in the concentrations of in alterations in the remodelling processes that take these enzymes in the aortic wall may change the place in the aortic wall and therefore the pattern of pattern of weakness leading to ‘favoured’ sites of aortic weakness. rupture.

Eur J Vasc Endovasc Surg Vol 18, August 1999 102 J. Golledge et al.

(a) Golledge et al. Darling et al.

14.5% 14%

73% 66%

12.5% 20%

(b) Golledge et al. Darling et al.

17%

35% 26%

29% 17% 37%

30% 9%

Fig. 3. Comparison of sites of aneurysm rupture demonstrated in this study and that of Darling and colleagues.11 Note the similar findings for the longitudinal site of rupture (a), but discordant findings regarding the transverse site (b).

Eur J Vasc Endovasc Surg Vol 18, August 1999 Morphology of Ruptured Abdominal Aortic Aneurysms 103

Many studies have demonstrated the disappointing stress and aortic wall tensile strength. Characteristics outcome following rupture of abdominal aortic an- of the patient and aneurysm may influence this eurysm.20,21 As can be seen from this study, most balance. The site that imbalance and, therefore, patients do not reach hospital, and even fewer the rupture occurs appears to influence the eventual operating theatre. Only 11 of the 46 patients (24%) outcome. studied underwent laparotomy, the majority (31, 67%) not reaching hospital or failing to respond to cardiopulmonary resuscitation on arrival. For those patients undergoing operation many factors have been References identified in determining outcome, including the se- verity of preoperative hypovolaemic shock, the age of 1Scott RA, Wilson NM, Ashton HA, Kay DN. Influence of the patient, comorbidity and the experience of the screening on the incidence of ruptured abdominal aortic an- 22–24 eurysm: 5 year results of a randomised controlled study. Br J surgeon. In our analysis we have assessed variables Surg 1995; 82: 1066–1070. predicting outcome for a group of patients with RAAA, 2Brown PM, Pattenden R, Vernooy C, Zelt DT, Gutelius which include those not reaching hospital or the op- JR. Selective management of abdominal aortic aneurysms in a prospective management programme. J Vasc Surg 1996; 23: erating theatre. The outcome for patients reaching 213–222. theatre was very favourable, with nine of 11 surviving. 3Humphrey JD, Kyriacou SK. The use of Laplace’s equation in Since many of the patients died prior to reaching aneurysm mechanics. Neurol Res 1996; 18: 204–208. 4Mower WR, Baraff LJ, Sneyd J. Stress distributions in vascular hospital the effect of some important factors, such as aneurysms: factors affecting risk of aneurysm rupture. J Surg hypotension, was impossible to assess. Surprisingly, Res 1993; 55: 155–161. 5Dobrin PB, Mrkvicka R. Failure of elastin or collagen as possible of the many patient variables assessed, the site of critical connective tissue alterations underlying aneurysmal di- aneurysm rupture was the most powerful predictor latation. Cardiovasc Surg 1994; 2: 484–488. of outcome (Table 4). Four of the nine patients who 6Anidjar S, Saltzmann JL, Gentrie D et al. Elastase-induced experimental aneurysms in rats. Circulation 1990; 82: 973–981. survived ruptured in the left posterior quadrant of 7Campa JS, Greenhalgh RM, Powell JT. Elastin degradation in the aneurysm, while none of the patients who died abdominal aortic aneurysms. Atherosclerosis 1987; 65: 13–21. ruptured at this site. The left anterior quadrant was 8Dobrin PB, Baker WH, Gley WC. Elastolytic and collagenolytic studies of arteries. Implications for the mechanical properties of the next safest site to rupture with two of 12 patients aneurysms. Arch Surg 1984; 119: 4405–4409. surviving (17%), while only one of the 16 patients (6%) 9Cronenwett JL, Murphy TF, Zelenock GB et al. Actuarial with right posterior quadrant ruptures survived. This analysis of variables associated with rupture of small abdominal aortic aneurysms. Surgery 1985; 98: 472–483. relative safety of left posterior rupture is difficult 10 Sterpetti AV, Cavallaro A, Cavallari N et al. Factors in- to explain and requires confirmation. We could only fluencing the rupture of abdominal aortic aneurysms. Surg Gyn- identify one other study which had investigated in ecol Obstet 1991; 173: 175–178. 11 D RC. Ruptured arteriosclerotic abdominal aortic an- 11 arling detail the site of abdominal aortic aneurysm rupture. eurysms. Am J Surg 1970; 119: 397–401. These authors, in studying PM specimens, found 12 Szilagyi DE, Elliot JP, Smith RF. Clinical fate of the patient with asymptomatic abdominal aortic aneurysm and unfit for left posterior quadrant ruptures to be as common surgical treatment. Arch Surg 1972; 104: 600–606. as right (Fig. 3). Operative cases were also in- 13 Cronenwett JL. Variables that affect the expansion rate and vestigated, but no indication was given as to the rupture of abdominal aortic aneurysms. Ann NY Acad Sci 1996; 800: 56–67. relative survival following rupture at different sites. 14 Katz DA, Littenberg B, Cronenwett JL. Management of small On anatomical grounds it might be suspected that abdominal aortic aneurysms. Early surgery versus watchful wait- the vertebral column and inferior vena cava would ing. JAMA 1992; 268: 2678–2686. 15 Stringfellow MM, Lawrence PF, Stringfellow RG. The in- provide greater protection for the right posterior fluence of aorta-aneurysm geometry upon stress in the aneurysm segment compared to the left posterior quadrant of wall. J Surg Res 1987; 42: 425–433. the aorta. One possible explanation for a worse 16 Elger DF, Blackketter DM, Budwig RS, Johansen KH. The influence of shape on the stresses in model abdominal aortic outcome following right posterior rupture might be aneurysms. J Biomech Engin 1996; 118: 326–332. that the relative protection at this site means that 17 Freestone T, Turner RJ, Coady A et al. Inflammation and matrix in order for a symptomatic blood loss to occur a metalloproteinases in the enlarging abdominal aortic aneurysm. Arterioscler Thromb Vasc Biol 1995; 15: 1145–1151. more significant rupture is required and, therefore, 18 Finlay GA, O’Driscoll LR, Russell KJ et al. Matrix metallo- the patients with ruptures at this site present later. proteinase expression and production by alveolar macrophages in emphsema. Am J Respir Crit Care Med 1997; 156: 240–247. In support of this hypothesis, the median time from 19 Liabakk NB, Talbot I, Smith RA, Wilkinson K, Balkwill F. symptoms to death was shortest for patients with Matrix metalloprotease 2 (MMP-2) and matrix metalloprotease right posterior ruptures (Fig. 2). 9 (MMP-9) type IV collaginases in colorectal cancer. Cancer Res 1996; 56: 190–196. In conclusion, this study supports the hypothesis 20 Armour RH. Survivors of ruptured abdominal aortic aneurysms: that rupture depends on a relationship between wall the iceberg’s tip. BMJ 1977; ii: 1055–1057.

Eur J Vasc Endovasc Surg Vol 18, August 1999 104 J. Golledge et al.

21 Johansson G, Swedenborg J. Ruptured abdominal aortic an- aortic aneurysm: impact of comorbidity and postoperative com- eurysms: a study of incidence and mortality. Br J Surg 1986; 73: plications on outcome. Ann Vasc Surg 1995; 9: 535–541. 101–103. 24 Rutledge R, Oller DW, Meyer AA, Johnson GJ. A statewide, 22 Cloutier L, Ricci MA, Wald S, Eldrup-Jorgensen J, Breden- population-based time-series analysis of the outcome of ruptured berg C. Predicting survival from ruptured abdominal aortic abdominal aortic aneurysm. Ann Surg 1996; 223: 492–505. aneurysm. Computer modelling with AIM versus clinical judge- ment. Ann NY Acad Sci 1996; 800: 234–235. 23 Panneton JM, Lassonde J, Laurendeau F. Ruptured abdominal Accepted 23 November 1998

Eur J Vasc Endovasc Surg Vol 18, August 1999