Evaluating Growth Patterns of Abdominal Aortic Aneurysm Diameter With Serial Computed Tomography Surveillance

Key Points

Question  What are the growth rates and patterns of small abdominal aortic aneurysms surveilled by computed tomography over 2 years?

Findings  In this cohort analysis of a randomized trial of 254 patients with 3.5- to 5.0-cm abdominal aortic aneurysms, average annual growth was 0.19 cm, and 70% of patients displayed linear growth.

Meaning  Based on linear low growth patterns, surveillance of abdominal aortic aneurysms less than 4.25 cm can be safely extended to at least 2 years.

Importance  Small abdominal aortic aneurysms (AAAs) are common in the elderly population. Their growth rates and patterns, which drive clinical surveillance, are widely disputed.

Objective  To assess the growth patterns and rates of AAAs as documented on serial computed tomography (CT) scans.

Design, Setting, and Participants  Cohort study and secondary analysis of the Non-Invasive Treatment of Abdominal Aortic Aneurysm Clinical Trial (N-TA³CT), a randomized, double-blind placebo-controlled clinical trial conducted from 2013 to 2018, with CT imaging every 6 months for 2 years. The trial was a multicenter, observational secondary analysis, not related to treatment hypotheses of data collected in the N-TA³CT. Participants included 254 patients with baseline AAA diameter between 3.5 and 5.0 cm.

Exposures  Patients received serial CT scan measurements, analyzed for maximum transverse diameter, at 6-month intervals.

Main Outcomes and Measures  The primary study outcome was AAA annual growth rate. Secondary analyses included characterizing AAA growth patterns, assessing likelihood of AAA diameter to exceed sex-specific intervention thresholds over 2 years.

Results  A total of 254 patients, 35 women with baseline AAA diameter 3.5 to 4.5 cm and 219 men with baseline diameter 3.5 to 5.0 cm, were included. Yearly growth rates of AAA diameters were a median of 0.17 cm/y (interquartile range [IQR], 0.16) and a mean (SD), 0.19 (0.14) cm/y. Ten percent of AAAs displayed minimal to no growth (<0.05 cm/y), 62% displayed low growth (0.05-0.25 cm/y), and 28% displayed high growth (>0.25 cm/y). Baseline AAA diameter accounted for 5.4% of variance of growth rate (P < .001; R², 0.054). Most AAAs displayed linear growth (70%); large variations in interval growth rates occurred infrequently (3% staccato growth and 4% exponential growth); and some patients’ growth patterns were not clearly classifiable (23% indeterminate). No patients with a maximum transverse diameter less than 4.25 cm exceeded sex-specific repair thresholds at 2 years (men, 0 of 92; 95% CI, 0.00-0.055; women, 0 of 25 ; 95% CI, 0.00-0.247). Twenty-six percent of patients with a maximum transverse diameter of at least 4.25 cm exceeded sex-specific repair thresholds at 2 years (n = 12 of 83 men with diameter ranging from 4.25 to <4.75 cm; 95% CI, 0.091-0.264; n = 21 of 44 men with diameter ranging from 4.75-5.0 cm; 95% CI, 0.362-0.669; n = 3 of 10 women with diameter ≥4.25 cm; 95% CI, 0.093-0.726).

Conclusions and Relevance  Most small AAAs showed linear growth; large intrapatient variations in interval growth rates were infrequently observed over 2 years. Linear growth modeling of AAAs in individual patients suggests smaller AAAs (<4.25 cm) can be followed up with a CT scan in at least 2 years with little chance of exceeding interventional thresholds.

Trial Registration  ClinicalTrials.gov Identifier: NCT01756833

Abdominal aortic aneurysm (AAA) affects 1.1 million people in the United States.¹ Most of these are small aneurysms without indication for immediate repair; they are managed expectantly with imaging surveillance until maximum transverse diameter (MTD) predicts a substantial enough risk of rupture to merit repair. Abdominal aortic aneurysm rupture or complication of repair were the primary cause of nearly 10 000 deaths in 2014 and a contributing cause of more than 17 000 deaths in 2009.^2,3 Ruptured AAA was the 17th leading cause of death in people older than 65 years in 2017.⁴ Understanding AAA growth patterns informs appropriate monitoring intervals. Quiz Ref IDDiscrepancies exist in the literature on the most common patterns of AAA growth. Some studies suggest staccato growth patterns^5,6; others suggest exponential growth^7-10; and others present growth as linear.^11,12

The Non-invasive Treatment of Abdominal Aortic Aneurysm Clinical Trial (N-TA³CT), a double-blind, placebo-controlled clinical trial of doxycycline randomized 261 participants with AAAs between 3.5 and 5.0 cm and measured the MTD of the AAA approximately every 6 months. The primary objective of this secondary cohort analysis is to describe growth rates and growth patterns of AAAs in this patient population. Secondarily, we seek to inform surveillance intervals by assessing likelihood of AAA diameter to exceed sex-specific intervention thresholds over 2 years and to clarify the association between baseline diameter and annual growth rates.

This study adhered to the Strengthening of Reporting of Observational Studies in Epidemiology (STROBE) statement and checklist (http://www.strobe-statement.org). Data in this study were collected from the N-TA³CT (n = 261); patients were enrolled from 22 US clinical centers between May 2013 and January 2017. The N-TA³CT enrolled patients 55 years or older with AAAs between 3.5 to 5.0 cm for men and 3.5 to 4.5 cm for women. Notable exclusion criteria included patients with AAAs less than 4.0 cm with documented failure to grow over the 2 years prior to enrollment, known connective tissue diseases, and current or planned treatment with systemic immunosuppressive agents. Institutional review boards at all clinical sites, core laboratories, and coordinating centers approved the protocol. Written consent was obtained from all patients. The study design, methods, and findings of N-TA³CT have been previously published.^13,14 The primary analysis indicated doxycycline did not affect aneurysm growth.¹⁴ Based on this finding and secondary analyses showing MTD growth rates (eFigure 1 in the Supplement) and patterns not differing significantly between treatment groups, Quiz Ref IDour analysis pooled both treatment arms.

Patients were included in our analysis if they had follow-up data beyond baseline imaging. Of the 261 patients enrolled in N-TA³CT, 254 had data allowing for their inclusion in the clinical outcomes analyses, 250 had 2 or more computed tomography (CT) scans enabling their inclusion in growth rate analyses, 241 had 3 or more CT scans enabling their inclusion in maximum segmental slope difference analysis, and 214 had 4 or more enabling their inclusion in growth pattern analyses (eFigure 2 in the Supplement).

Abdominal and pelvic CT scans were obtained at baseline and approximately every 6 months thereafter for 18 to 30 months. A total of 204 patients (80%) completed imaging through the 24-month window; median period of follow-up was 26 months (interquartile range [IQR], 24-31 months). The AAA MTD was assessed via CT imaging processed at AortaCore Imaging Core Laboratory at the University of Wisconsin, Madison on postprocessing workstations (Aquarius Intuition, version 4.4.12; TeraRecon). The protocol included masking of patient identity, digital magnification, standardized windowing, and quality control with blind replicates and twice-yearly proficiency testing. A single reader, with many years of aortic CT analysis experience, used the double oblique technique to measure the MTD; intraobserver and interobserver intraclass correlation coefficients were greater than 0.99 for this technique.¹³ The double oblique technique selects the multiplanar reconstruction image of the largest section of the AAA, following a manually corrected centerline, in a plane perpendicular to the aneurysm walls. The largest diameter of the aneurysm was measured with electronic calipers from outer wall to outer wall. This study used the double oblique method to measure MTD because it is more reproducible and accurate than measuring diameters on axial CT or ultrasonography images.^15,16

Each patient’s diameter growth rate was estimated by simple linear regression of MTD on time using all available CT scans performed in the study; growth rate (slope of linear regression) was expressed as centimeters of growth per year. Annual growth rates were grouped into 3 categories: minimal to no growth (<0.05 cm/y), low growth (0.05-0.25 cm/y), or high growth (>0.25 cm/y). These categories were established a posteriori based on (1) ±0.05 cm as an acceptable difference for reproducibility of AAA diameter measurements on CT scans^17,18 and (2) data from prior surveillance studies of AAAs comparable with N-TA³CT, which reported mean annual growth rates ranging from 0.19 cm¹⁹ to 0.28 cm.²⁰ The Shapiro-Wilk test was used to assess whether distribution of growth rates was approximately normal.²¹

Besides the simple linear regression modeling of MTD on time (linear model), regressions of MTD on e^time (exponential model) were performed. Linear growth was defined as an R² value of at least 0.90 for the linear model. Patients with nonlinear growth were subcategorized into exponential growth, staccato growth, or indeterminate growth patterns. Exponential growth was defined by an R² value of at least 0.95 for the exponential model. Staccato growth pattern was defined as a pattern of 1 or more “no significant growth intervals” (≤0.05 cm/interval) accompanied by 1 or more “fast growth intervals” (≥0.6 cm/y).⁵ Growth patterns that did not fit the definitions here were classified as indeterminate and further categorized as indeterminate–not growing (annual growth rate <0.05 cm/y), and indeterminate-growing (≥0.05 cm/y). Two investigators (S.O. and J.M.), individually and then jointly, reviewed every individual patient’s growth pattern to qualitatively confirm that calculated growth rates and pattern assessments were appropriate; no growth patterns were changed on review.

The observed growth rate between each successive pair of measurements was calculated and the range (maximum minus minimum) of sequential growth rate differences, referred to here as the maximum segmental slope difference (max-SSD), was also calculated. Perfect, undeviating linear growth would result in a max-SSD of 0. Thus, max-SSD was used to further describe linear growth, and was calculated for each patient for whom there were at least 3 images (n = 241). The Shapiro-Wilk test was used to assess whether distribution of max-SSD was approximately normal.²¹

Association of growth rate, calculated as described in previous sections, with baseline aneurysm diameter was assessed by simple linear regression of a patient’s growth rate based on their baseline MTD.

Probabilities of meeting the MTD threshold for repair (5.5 cm for men and 5.0 cm for women) within 2 years or having AAA repair within 2 years were estimated by the life table method to include data for patients with less than 2 years of follow-up. Confidence intervals were calculated using exact binomial calculation for proportions. Results with P at or less than .05 were considered statistically significant, and all P values were 2-sided. No correction for multiple comparisons was made. Statistical analysis was performed using SAS, version 9.4 (SAS Institute and NCSS 10 and NCSS 2020 (Number Cruncher Statistical Systems).

The N-TA³CT randomization occurred between May 2013 and January 2017. Of 261 patients randomized, 254 had follow-up beyond baseline that could be used in these analyses. The baseline characteristics of the 254 patients are summarized in Table 1. The mean (SD) age of patients was 70.9 (7.4) years and 86% (n = 219) were men. Ninety-two percent (n = 235) were current or former smokers. Comorbidities included hypercholesterolemia (n = 196; 77%), coronary artery disease (n = 104; 41%), cancer (n = 82; 32%), and chronic obstructive pulmonary disease (n = 59; 23%). Baseline AAA MTDs ranged from 3.50 to 4.52 cm for women and 3.50 to 5.04 cm for men.

Distribution of yearly growth rates shows mean growth of 0.19 cm/y and median growth of 0.17 cm/y (IQR, 0.16) (Figure 1). The Shapiro-Wilk test showed AAA MTD annual growth rate to be not normally distributed, with a right skew. Quiz Ref IDAnnual growth rates in the 3 assigned groups were: minimal to no growth (<0.05 cm/y), 10% of patients (n = 25); low growth (0.05-0.25cm/y), 62% (n = 156); and high growth (>0.25 cm/y), 28% (n = 69) (eFigure 3 in the Supplement). eFigure 4 in the Supplement displays 3 individual patient examples that are representative of growth rates of minimal to no growth (eFigure 3A), low growth (eFigure 3B), and high growth (eFigure 3C). On linear regression, 10 patients had negative slopes (eFigure 5 in the Supplement) consistent in absolute magnitude with minimal to no growth or low growth.

Maximum segment slope difference ranged from negligible to 1.6 cm/y, with a mean (SD) of 0.27 (0.18) cm/y; median, 0.23 (IQR, 0.16) (Figure 2). The Shapiro-Wilk test for normal distribution showed max-SSD to be not normally distributed; there is a right skew. Quiz Ref IDThe small differences between slope segments in most patients support the linear model fit of MTD growth rates per patient over 2 years. eFigure 6 in the Supplement displays 3 individual patient examples, with corresponding max-SSDs, that are representing undeviating linear growth, linear growth with some variation over time, and nonlinear growth. The distribution of growth patterns was not significantly different between N-TA³CT treatment groups (placebo [n = 107]: 67% linear, 5% exponential, 6% staccato, 11% indeterminate-growing, and 11% indeterminate–not growing; doxycycline [n = 107]: 72% linear, 3% exponential, 1% staccato, 8% indeterminate-growing, and 16% indeterminate–not growing).

Linear growth was observed in 70% of patients (95% CI, 0.630-0.757) assessed for growth pattern (16.1% of patients with 4 scans, 32.2% with 5 scans, and 51.7% with 6 scans). Nonlinear growth patterns were infrequent: staccato was observed in 7 patients (3%; 95% CI, 0.03-0.066) (14.3% of patients with 4 scans, 57.1% with 5 scans, and 28.6% with 6 scans) and exponential in 8 patients (4%; 95% CI, 0.016-0.072) (87.5% of patients with 5 scans and 12.5% with 6 scans). Twenty-three percent of patients had indeterminate growth patterns: 29 patients with indeterminate-growing (13%; 95% CI, 0.093-0.189) (20.7% of patients with 4 scans, 41.4% with 5 scans, and 37.9% with 6 scans) and 21 patients with indeterminate–not growing (10%; 95% CI, 0.062-0.146) patterns (19.0% of patients with 4 scans, 19.0% with 5 scans, and 61.9% with 6 scans). Patients with linear growth had lower max-SSDs than patients with nonlinear growth (eFigure 7 in the Supplement). eFigure 8 in the Supplement displays 3 individual patient examples representing nonlinear growth patterns of staccato growth (eFigure 8A), exponential growth (eFigure 8B), and indeterminate growing (eFigure 8C). Applying a previously published staccato growth definition, we identified 22 patients (10%) with staccato growth.⁵

Pretreatment AAA MTD was significantly associated with yearly MTD growth rate but accounted for only 5.4% of the variance of growth rate (Figure 3). A 1-cm increase in baseline diameter was associated with 0.08-cm/y increase in annual growth rate (95% CI, 0.04-0.11). Minimal to no growth and high growth rates were observed across the range of baseline MTDs. For patients with baseline MTD less than 4.25 cm, 0% exceeded surgical repair thresholds. Among men whose baseline MTD was less than 4.25 cm, the 95% CI upper limit to the frequency of reaching threshold over 2 years was 5.5%. Specific frequencies of surpassing repair thresholds are given for each size range in Table 2. No patients in N-TA³CT had ruptured AAAs. Of the 24 patients repaired over the study, only 4 exceeded the MTD threshold for repair on the CT imaging preceding the repair.

These data demonstrate that most small AAAs exhibit low growth in a linear pattern. The annual growth rate of AAA diameter was lower than anticipated, with a mean of 0.19 cm/y and median of 0.17 cm/y.¹³ A small proportion of aneurysms with high growth rates gives the annual growth rates a nonnormal distribution with a right skew, a finding seen in a prior retrospective analysis of AAAs.^22,23

Ten percent of patients displayed minimal to no growth over 2 years. A surveillance study by Thompson et al²⁰ of small AAAs in England (median baseline size 3.5 cm, IQR 3.1-4.2) reported a bimodal distribution of growth with around half of AAAs remaining quiescent with little growth and the other half continuing to expand. A similar surveillance study of 3.0- to 3.9-cm aneurysms from the US Veteran Affairs system reported 25% of aneurysms did not grow at all over a median follow-up of 3.5 years.²³ These studies’ findings raise questions about whether the aortas with small, nongrowing MTDs are truly aneurysmal or have a different etiology/pathophysiology.²⁰ The N-TA³CT design excluded patients with AAA diameters less than 4.0 cm who had documented failure of the aneurysm to grow over the 2 years prior to enrollment, which could result in this study underestimating the proportion of aneurysms that are stable or very slow growers. Likely, the true proportion of small AAAs that do not grow lies somewhere between our reported 10% no growth and the Thompson et al 50%.²⁰

Ten patients exhibited small “negative” growth rates. These patients have stable nonshrinking aneurysms (eFigure 5 in the Supplement); small biologic fluctuations (cardiac cycle, blood pressure, and hydration),²⁴ measurement variability, and serial surveillance captured a trivial regression from baseline imaging to a stable mean diameter. While drug interventions that regress preformed AAAs in animal models are an active field of study,^25,26 spontaneous regression of human AAAs has been reported in only a small number of cases.²⁷

The data support an association between initial MTD and growth rate, although the effect size is small. Over 2 years, we observed growth rates with “minimal to no growth” and “high growth” over the range of baseline MTDs. Based on these 2-year follow-up data, our study supports prior findings that baseline MTD is not a strong predictor of future growth of small AAAs.^11,15,20 Importantly, the association of a cross-sectional measure of MTD at baseline with growth rate should not be taken as a basis for inferring exponential growth. To support an exponential growth pathophysiology, growth within a patient proportional to diameter over time would have to be observed.

Quiz Ref IDFor male patients with small AAAs who meet the enrollment criteria of N-TA³CT, we conclude that a small AAA (3.5-4.24 cm) can be monitored at a frequency of at least 2 years, with little chance of exceeding the interventional MTD threshold (5.5 cm). Our data on women are lesser in amount and cannot be used to support a definitive conclusion on their own, but they are consistent with the data on men. Our findings are consistent with the RESCAN collaboration, a collection of 15 471 patients from 18 clinical data sets with repeated ultrasonography measurements, which concluded that surveillance intervals for AAA could be safely extended.²⁸ The present study capitalized on the detailed and precise serial CT imaging to ascertain patterns of growth, which are predominately linear and predictable. Our recommendation for extending imaging intervals based on the linearity and predictability of AAA growth measured via CT scans supports the conclusions of RESCAN based on longer-term ultrasonography surveillance and clinical outcomes.

The finding that 70% of AAAs undergo linear and predictable growth behavior has 3 clinical implications. First, patients with early-stage disease may be reassured that their disease is unlikely to suddenly alter course. Second, along with others’ findings,²⁰ eg, the RESCAN collaboration,^10,28 recommended intervals for surveillance²⁹ could be extended safely to reduce costly and unnecessary CT scan imaging. Third, monitoring of small AAAs is safe because sudden unpredictable AAA growth is infrequent.^30-32 Of 116 patients with baseline AAA MTD less than 4.25 cm, none progressed to repair thresholds at 2 years. Obtaining CT images at 2-year intervals is reasonable. Additionally, the N-TA³CT findings¹⁴ are consistent with 4 randomized trials (UKSAT,³³ ADAM,³⁰ PIVOTAL,³⁴ and CAESAR¹⁹) that found no benefit for intervention on small aneurysms and are discordant with the suggestion from a epidemiologic study,³⁵ vulnerable to the ecological fallacy, that practice guidelines should be reevaluated in favor of indicating earlier intervention.³⁵

This study adds to and can clarify the different perspectives on AAA growth patterns. The max-SSD statistic identifies AAAs with nonlinear growth and displays a right skew in its distribution. Most patients display linear growth, with few patients showing markedly nonlinear patterns of growth that could be described as staccato or exponential. A 2004 study by Kurvers et al⁵ reported staccato growth in 65% of 52 patients with AAAs who were too high risk for open repair and unsuitable for endovascular repair. That study and others reporting large frequencies of staccato growth compare unequal time intervals between images, which could artificially create the appearance of growth arrest and/or high growth intervals.^5,36 Those studies have suggested that aneurysm growth is unpredictable and shows large variations over time. Applying the Kurvers definition to the patients in this study, 22 patients (10%) qualified for staccato growth, although 15 did not appear staccato on review; 5 of these 15 demonstrated high growth over 12 months or longer and 10 were classified as staccato on the basis of a single measurement deviating from a linear pattern making the analysis vulnerable to random biologic variation (eg, owing to changes in blood pressure) and CT scan performance/interpretation.^37-39 Using a definition of staccato growth with consistent time intervals between CT scans, there were 7 patients with staccato growth patterns (3%) in our data. There were 8 patients (4%) whose growth might be better described as exponential rather than linear. The patients with “minimal to no growth” rates were classified as having an “indeterminate–not growing” pattern because, although there was no significant growth over the 2 years we observed, at some point these aneurysms must have been developing and growing, and therefore we are unable to identify a clear growth pattern with measurements for only 2 years.

The aneurysm diameter growth pattern categories we observed inform 2 major conclusions and future directions. First, longer-term follow-up is necessary to confirm growth patterns and trends over the course of aneurysmal disease that may not be apparent within 2 years of surveillance. Second, although nonlinear growth is infrequent, it could be clinically important and justifies investigation of biomarkers to predict future growth or intervals without growth; such studies could provide information to shape future therapeutic approaches for small aneurysms.

This study is limited by its sparse follow-up beyond 2 years; growth patterns or variability that manifest more clearly over longer periods may have been missed. Additionally, this study analyzed data of patients with AAAs 5.0 cm or less in diameter at baseline; AAAs with diameters greater than 5.0 cm may demonstrate different growth patterns. As previously noted, eligibility criteria of N-TA³CT limited the inclusion of patients and also insisted on well-controlled blood pressure and encouraged smoking cessation. A greater proportion of patients with diabetes (22%) than enrolled in previous study populations could have altered growth trends in the direction of more predictable slow growth. The small number of female patients (35; 14%) reduces the precision of our observations about AAA growth among women, particularly evident in subgroup analyses. Fast-growing AAAs may be slightly underrepresented in our data based on accelerated progression reducing the window in which they would meet MTD eligibility criteria for N-TA³CT and because faster growth rates lead surgeons to select out patients for early repair. Comprehensive address of confounders or predictors of growth rates are beyond the scope of this article and are important future directions.

Most small AAAs (3.5-5.0 cm) exhibit “minimal to no” or “low” yearly growth (0.05-0.25 cm/y). Growth is usually linear. Great variability in growth over a 2-year period is infrequent. Linear growth modeling of AAAs in individual patients suggests small AAAs (<4.25 cm) can be followed up with a CT scan 2 years later, with a small chance of exceeding interventional MTD thresholds of 5.5 cm for men.

Corresponding Author: Sydney L. Olson, BS, Department of Vascular Surgery, University of Wisconsin-Madison, 600 Highland Ave, H4/635, Madison, WI 53792 (olsons@surgery.wisc.edu).

Accepted for Publication: December 8, 2020.

Published Online: February 17, 2021. doi:10.1001/jamasurg.2020.7190

Author Contributions: Ms Olson and Dr Terrin had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Olson, Wijesinha, Panthofer, Upchurch, Terrin, Curci, Matsumura.

Acquisition, analysis, or interpretation of data: Olson, Wijesinha, Panthofer, Blackwelder, Terrin, Curci, Baxter, Matsumura.

Drafting of the manuscript: Olson, Panthofer, Baxter.

Critical revision of the manuscript for important intellectual content: Olson, Wijesinha, Panthofer, Blackwelder, Upchurch, Terrin, Curci, Matsumura.

Statistical analysis: Olson, Wijesinha, Blackwelder, Terrin.

Obtained funding: Upchurch, Terrin, Baxter.

Administrative, technical, or material support: Olson, Panthofer, Terrin, Matsumura.

Supervision: Terrin, Curci, Baxter, Matsumura.

Conflict of Interest Disclosures: Dr Blackwelder reported receiving grant support from the National Institutes of Health (NIH). Dr Terrin reported receiving grant support from the NIH. Dr Curci reported receiving grant support from the NIH. Dr Baxter reported receiving grant support from the National Institute on Aging (NIA-NIH). Dr Matsumura reported receiving grant support from the NIH, Abbott, Cook, Endologix, Gore, and Medtronic. No other disclosures were reported relative to this work.

Funding/Support: This work was supported by grants R34 AG028684, R01 AG037120, T32 AG000262, and P30 AG0288747 from the National Institute on Aging. The N-TA³CT was funded by the National Institutes of Health through a grant award to the University of Maryland, Baltimore.

Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the clinical trial, collection, management, or analysis of the data or interpretation of the results. The National Institutes of Health appointed an independent data and safety monitoring board. Neither the National Institutes of Health nor the data and safety monitoring board had any role in the preparation, review or approval of this manuscript or the decision to submit it for publication.

Additional Contributions: We thank Jennifer Grudzinski, BS, RT(R)(M) (University of Wisconsin School of Medicine and Public Health), for completing the rigorous measurements that made this project possible.