A BMI-adjusted ultra-low-dose CT angiography protocol for the peripheral arteries—Image quality, diagnostic accuracy and radiation exposure

European Journal of Radiology 93 (2017) 149–156

Contents lists available at ScienceDirect

European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad

Original Article

A BMI-adjusted ultra-low-dose CT angiography protocol for the peripheral arteries—Image quality, diagnostic accuracy and radiation exposure

MARK

Markus M. Schreinera, Hannes Platzgummera, Sylvia Unterhumera, Michael Webera, ⁎ Gabriel Mistelbauerb, Christian Loewea, Ruediger E. Schernthanera, a Section of Cardiovascular and Interventional Radiology, Department of Biomedical Imaging and Image-Guided Therapy, Medical University of Vienna, Waehringer Guertel 18-20, 1090 Vienna, Austria b Institute of Computer Graphics and Algorithms, Technical University of Vienna, Favoritenstraße 9-11, 1040 Vienna, Austria

A R T I C L E I N F O

A B S T R A C T

Keywords: Peripheral arterial disease CT angiography Digital subtraction angiography Radiation exposure Body mass index

Objectives: To investigate radiation exposure, objective image quality, and the diagnostic accuracy of a BMIadjusted ultra-low-dose CT angiography (CTA) protocol for the assessment of peripheral arterial disease (PAD), with digital subtraction angiography (DSA) as the standard of reference. Methods: In this prospective, IRB-approved study, 40 PAD patients (30 male, mean age 72 years) underwent CTA on a dual-source CT scanner at 80 kV tube voltage. The reference amplitude for tube current modulation was personalized based on the body mass index (BMI) with 120 mAs for [BMI ≤ 25] or 150 mAs for [25 < BMI ≤ 30]. Iterative image reconstruction was applied. The presence of significant stenoses (> 70%) was assessed by two readers independently and compared to subsequent DSA. Radiation exposure was assessed with the computed tomography dose index (CTDIvol) and the dosis-length product (DLP). Objective image quality was assessed via contrast- and signal-to-noise ratio (CNR and SNR) measurements. Radiation exposure and image quality were compared between the BMI groups and between the BMI-adjusted ultra-low-dose protocol and the low-dose institutional standard protocol (ISP). Results: The BMI-adjusted ultra-low-dose protocol reached high diagnostic accuracy values of 94% for Reader 1 and 93% for Reader 2. Moreover, in comparison to the ISP, it showed significantly (p < 0.001) lower CTDIvol (1.97 ± 0.55 mGy vs. 4.18 ± 0.62 mGy) and DLP (256 ± 81 mGy x cm vs. 544 ± 83 mGy x cm) but similar image quality (p = 0.37 for CNR). Furthermore, image quality was similar between BMI groups (p = 0.86 for CNR). Conclusions: A CT protocol that incorporates low kV settings with a personalized (BMI-adjusted) reference amplitude for tube current modulation and iterative reconstruction enables very low radiation exposure CTA, while maintaining good image quality and high diagnostic accuracy in the assessment of PAD.

1. Introduction Peripheral arterial disease (PAD) is an important health issue, with increasing incidence rates due to the demographic patterns of the last decades [1,2]. Although PAD is usually diagnosed clinically, a complete radiological assessment, including the in- and outflow, is mandatory for optimal treatment planning [3]. Because digital subtraction angiography (DSA) provides the highest spatial and temporal resolution, it is still considered the standard of reference, despite inherent limitations [4]. Driven by technological advances in the last two decades, computed tomography angiography (CTA) has evolved into an accurate

⁎

[5–7] and cost-efficient [8] imaging alternative for patients with intermittent claudication [9] or critical limb ischemia [10]. However, CTA has several limitations. For example, severe vessel calcifications cause blooming artifacts which in return compromise the confidence in the accuracy of the modality for the detection of hemodynamically significant stenoses [11]. In addition, the evaluation of a CTA based exclusively on axial images is rather time-consuming and prone to errors [12,13]. Thus, additional post-processing is necessary to facilitate the evaluation by means of three-dimensional reformations. Despite the introduction of dual-energy CT scanners and their advanced capabilities to discriminate iodine and calcium, full-automatic post-processing is

Corresponding author. E-mail addresses: [email protected] (M.M. Schreiner), [email protected] (H. Platzgummer), [email protected] (S. Unterhumer), [email protected] (M. Weber), [email protected] (G. Mistelbauer), [email protected] (C. Loewe), [email protected] (R.E. Schernthaner). http://dx.doi.org/10.1016/j.ejrad.2017.06.002 Received 4 October 2016; Received in revised form 13 May 2017; Accepted 1 June 2017 0720-048X/ © 2017 Elsevier B.V. All rights reserved.

European Journal of Radiology 93 (2017) 149–156

M.M. Schreiner et al.

not flawless yet [14,15], so that manual, time-consuming interaction is still required [16]. The main drawbacks however are the necessity for contrast media administration and radiation exposure. In particular, when considering that increasing life expectancy and steadily improving treatment options require an incremental number of repetitive assessments in the course of the disease, resulting in a considerable cumulative exposure to both contrast agents and ionizing radiation [17], the former carrying the risk of contrast-induced kidney injury and the latter the risk of cancerogenesis [18]. Options to reduce radiation exposure include the reduction of tube voltage and modulation of the tube current [19]. The inevitable increase in image noise associated with the reduction in ionizing radiation might be compensated by the additional application of iterative image reconstruction [20]. Some of these techniques have already been applied to the peripheral arteries [21,22]. However, to the best of our knowledge, a comprehensive integration of these techniques in one imaging protocol, with special consideration of a patient’s body mass index (BMI), has not yet been evaluated. The BMI is of particular importance as it influences radiation absorption, and thus, image quality [23], for peripheral CTA especially affecting image quality in the region of the iliac arteries. We put forward the conjecture that additional adaptation of the tube current based on BMI would reduce radiation exposure even further while maintaining a high image quality. The purpose of this study was to investigate radiation exposure, image quality, and the diagnostic accuracy of a personalized BMI-adjusted ultra-low-dose CTA protocol for the assessment of PAD, with DSA as the standard of reference.

another four-second delay, a scan was initiated that covered the entire volume from the renal arteries to the mid-foot. The scan time was set to 40 s with a pitch of 0.4, and the table increment was adjusted accordingly. Although the tube voltage was set to 80 kV in all patients, the reference for tube current modulation was adjusted for the BMI. Group 1 (BMI ≤ 25) was scanned with a reference tube current of 120 mAs, whereas 150 mAs was selected for group 2 (25 < BMI ≤ 30). Then, 1.5 mm thick transverse sections were reconstructed at 1 mm intervals, using an I30f kernel and sinogram-affirmed iterative reconstruction (SAFIRE strength level 3, Siemens Medical Systems, Erlangen, Germany). Based on the known expansion of signal and noise in projection data, SAFIRE iteratively determines the noise content, subtracts it from the image data-set and validates it via comparison to the initial raw data. Each iterative step hence reduces noise further, but at the cost of increased image smoothing [25–27]. For subsequent evaluation, multipath curved planar reformations (mpCPRs) were created over a viewing range of 180° (from right lateral [–90°] through anteroposterior [0°] to left lateral [+90°] viewing angles) in 9° intervals and saved in DICOM format to preserve the original CT resolution and attenuation information using a semi-automated toolbox (Supplementary DICOM dataset) [16]. 2.2. Digital subtraction angiography DSA was performed routinely during endovascular therapy on an Axiom Artis, Angiostar or an ArtisZeego Digital Angiography System (Siemens Systems, Erlangen, Germany) via antegrade or retrograde puncture of a common femoral artery. A low osmolar, non-ionic, iodinated contrast agent (Ioversol, Optiray 350, Covidien, Austria) was used, which contained 350 mg iodine per ml.

2. Materials and methods This prospective, single-center study was approved by the local institutional review board and was conducted according to the Declaration of Helsinki, including current revisions. Written, informed consent was obtained from all patients prior to recruitment. Based on an a priori power analysis we aimed to prospectively include 40 patients referred for CTA of the peripheral arteries for PAD treatment planning (stage II–IV, according to Fontaine’s classification). Specialists in internal medicine assessed the PAD diagnosis and the Fontaine’s stage. The following served as inclusion criteria: age older than 18 years; BMI below 30; estimated glomerular filtration rate above 30 mL/ min; and normal TSH levels. Pregnancy and breastfeeding were defined as exclusion criteria. Depending on the BMI, patients were assigned to two different groups (group 1: BMI ≤ 25, n = 25; group 2: 25 < BMI ≤ 30, n = 15). Patients who had been treated conservatively or with surgical revascularization had to be excluded from the study for lack of the standard of reference (DSA). Patients, who were treated by endovascular therapy more than 30 days after the CTA, were excluded as well to avoid a bias attributable to disease progression between the CTA and the intra-procedural DSA.

2.3. Objective image quality and radiation exposure All patient data were kept confidential and were evaluated in a blinded fashion. An expert vascular radiologist (Reader 1, R.E.S., ten years of experience) assessed objective image quality, based on contrast-to-noise ratio (CNR) and signal-to-noise-ratio (SNR) on a picture archiving and communication system (PACS) workstation (IMPAX EE R20, Agfa Healthcare N.V., Mortsel, Belgium). In each patient, seven vessel regions (aorta, right and left iliac bifurcation, right and left femoral bifurcation, right and left tibial peroneal trunk) were evaluated via regions-of-interest (ROI) analysis using the Hounsfield Unit (HU) scale. Special care was taken to avoid calcified lesions or artifact-related structures and to ensure that each ROI covered the entire vascular lumen. The following muscles were used for reference measurements on the same axial image as the vessel measurement: the psoas muscle; the biceps femoris muscle; and the soleus muscle. CNR and SNR were then calculated according to:

CNR =

2.1. Multidetector CT angiography

SNR = All CTA studies were conducted on a second-generation, dualsource, multi-detector CT scanner (Somatom Definition Flash, Siemens Medical Systems, Erlangen, Germany) with a detector configuration of 128 × 0.6 mm. Ninety ml of a low-osmolar, non-ionic iodinated contrast agent (Ioversol, Optiray 350, Covidien, Austria) was administered with a programmable power injector using a dedicated injection protocol (OptiBolus, Covidien, Austria). The protocol consisted of a monophasic contrast agent injection with an exponentially decreasing flow rate (3.5–2.6 ml/s) over 35 s, followed by a 35 ml saline flush at a flow rate of 2.6 ml/s [24]. Ten seconds after starting the contrast injection, the first reference scan was performed in the aorta at the level of the origins of the renal arteries and was repeated each second until the aorta exhibited an enhancement of 150 Hounsfield units (HU). After

meanHUvessel − meanHUmuscle SDmuscle

meanHUvessel SDmuscle

CNR and SNR values were used to compare objective image quality between the two BMI groups of the BMI-adjusted ultra-low-dose protocol. Furthermore, we identified all patients of our study cohort who had previously undergone CTA on the same scanner with the low-dose institutional standard protocol and retrieved these scans along with DLP and CTDIvol measurements. These scans were evaluated as described above as well and used to intra-individually compare both, objective image quality (SNR and CNR) and radiation exposure (DLP and CTDIvol) between the BMI-adjusted ultra-low-dose protocol and the low-dose institutional standard protocol. The tube voltage of the low-dose institutional standard protocol was 80 kV as well, but a fixed reference amplitude for tube current 150

European Journal of Radiology 93 (2017) 149–156

M.M. Schreiner et al.

Fig. 1. A 72-year-old male patient with a BMI of 27.5 was referred to interventional radiology with PAD stage IIb. The arterial tree of each leg was divided into 21 segments, as demonstrated on mpCPR at a viewing angle of 45° (left-oblique view): CIA = common iliac artery, EIA = external iliac artery, IIA = internal iliac artery, CFA = common femoral artery, DFA = deep femoral artery, SFA = superficial femoral artery, POP = popliteal artery, ATA = anterior tibial artery, TPT = tibioperoneal trunk, PA = peroneal artery, PTA = posterior tibial artery. SFA, POP, ATA, PA, and PTA were subdivided into three segments: proximal, middle, and distal.

modulation of 285 mAs was used. In addition, image reconstruction was based on filtered-back projections only and no iterative reconstruction algorithm was applied. 2.4. Diagnostic performance For the evaluation of diagnostic performance, the arterial tree of each leg was divided into three vessel territories (iliac, femoro-popliteal and infra-popliteal arteries) and 21 vessel segments (Fig. 1). Two readers, one expert vascular radiologist (Reader 1, R.E.S., ten years of experience) and one radiological specialist in training (Reader 2, H.P., five years of experience) independently assessed CTA based on axial images and mpCPRs. DSA images, which served as the standard of reference, were assessed exclusively by the senior radiologist (Reader 1). To eliminate recall bias, the readings of DSA and CTA images were separated by eight weeks. For each segment, the most severe stenosis was assessed for hemodynamic significance (> 70 %). Five additional categories were defined for segments that were not assessable due to one of the following reasons: not depicted; insufficient contrast; severe calcifications; prosthesis-related artifacts; or stent-related artifacts. 2.5. Statistical analysis All statistical evaluations were performed using IBM SPSS for Windows, Version 22.0.0.2 (IBM, NY, USA). Nominal data (e.g., ratings) were described using absolute frequencies and percentages. Metric data (e.g., CNR) were described using mean ± standard deviation given normal distribution, or median, first, and third quartile in case of skewed data. An a priori power analysis was performed to estimate the required sample size for the accuracy assessment based on previous studies at our site. The diagnostic performance of CTA was assessed using logistic regression for repeated measures with DSA as the standard of reference. For inter-reader agreement, kappa values were calculated and interpreted according to the criteria of Landis and Koch [28]. For intra-individual comparison of CNR, SNR, CTDIvol and DLP between the institutional standard and the personalized protocol, paired t-tests were performed. For inter-individual comparison of CNR, SNR, CTDIvol and DLP between the two BMI sub-groups, unpaired ttests were performed. Normal distribution was ensured at all times via a Kolmogorov-Smirnov-test. A p-value less than 0.05 was considered significant. 3. Results Between September 2012 and April 2014, 183 consecutive patients fulfilled the inclusion criteria. 143 patients had to be excluded because endovascular therapy was not performed at all or not within 30 days of the CTA (Fig. 2 - Flowchart ). Of the 40 patients who were prospectively included in this study, 30 were male and 10 were female (mean age, 72 ± 11 years; range, 44–101). Six patients had a total of eight orthopedic implants, and 17 patients had a total of 47 vascular stents. Additional demographic characteristics, including risk factors for PAD, are displayed in Table 1. The average interval between CTA and subsequent DSA was six days, with a range of 1-28 days. In 31 patients, endovascular therapy was limited to one leg. In 16 of those patients, DSA was performed after antegrade arterial puncture of the common femoral artery, thus depicting only segments downriver of the puncture 151

European Journal of Radiology 93 (2017) 149–156

M.M. Schreiner et al.

Fig. 2. Flow-chart illustrating patient recruitment and the allocation to sub-cohorts.

Table 1 Patient Characteristics. Characteristics

Age (years)

*

Sex Male Female Weight (kg)

*

Height (cm)

*

BMI* Fontaine Stage IIb III IV Risk factors Smoker Hyperlipidemia Hypertension Diabetes Coronary artery disease

Table 2 Inter-individual comparison of the signal-to-noise-ratios (SNR) and contrast-to-noise-ratios (CNR) on axial images between group 1 (BMI ≤ 25) and group 2 (25 < BMI ≤ 30). Group 1 (n = 25)

Group 2 (n = 15)

Overall (n = 40)

72 ± 12 (52–101)

72 ± 10 (44–87)

72 ± 11 (44–101)

15 (60) 10 (40)

15 (100) 0 (0)

30 (75) 10 (25)

63.2 ± 9.7 (44.0−81.0) 169 ± 9 (156–184) 22.1 ± 2.4 (14.2–25.0)

82.6 ± 9.1 (66.0−93.0) 173 ± 7 (162–184) 27.2 ± 1.5 (25.1–29.7)

70.5 ± 13.3 (44.0−93.0) 171 ± 8.5 (156–184) 24.0 ± 3.3 (14.2–29.7)

10 (40) 2 (8) 13 (52)

10 (67) 1 (7) 4 (26)

20 (50) 3 (8) 17 (42)

12 (48) 12 (48) 14 (56) 11 (44) 6 (24)

5 8 7 5 4

17 20 21 16 10

Vessel regions

Group 1 (n = 25)

Group 2 (n = 15)

p-value

CNR

Iliac bifurcation Femoral bifurcation Tibial peroneal trunk All regions

9.67 ± 3.38 22.58 ± 8.40 26.42 ± 10.25 19.28 ± 10.60

8.86 ± 3.10 20.77 ± 9.25 30.55 ± 13.61 19.57 ± 12.84

0.29 0.37 0.15 0.86

SNR

Iliac bifurcation Femoral bifurcation Tibial peroneal trunk All regions

10.87 25.17 29.01 21.43

10.00 23.32 34.12 21.94

0.30 0.41 0.12 0.77

± ± ± ±

3.70 9.03 11.90 11.75

± ± ± ±

3.43 10.25 14.98 14.21

All data are given as mean ± standard deviation.

3.1. Objective image quality

(33) (53) (47) (33) (27)

3.1.1. Inter-individual comparison of objective image quality between BMI groups No significant differences in overall CNR (p = 0.86) and overall SNR (p = 0.77) were found between group 1 and group 2 of the study cohort (Table 2), which is also depicted visually in Fig. 3.

(43) (50) (53) (40) (25)

3.1.2. Intra-individual comparison of objective image quality between the low-dose ISP and the BMI-adjusted ultra-low-dose protocol We identified 15 patients, who had previously undergone assessment of the peripheral arteries on the same scanner with the low-dose ISP. Overall, no significant differences in CNR (p = 0.37) and SNR (p = 0.23) were found between the BMI-adjusted low-dose protocol and low-dose ISP. In particular, the CNR was similar for the vessels of the thigh (p = 0.80) and the lower leg (p = 0.79). Merely in the pelvic region, the CNR of the personalized protocol was significantly lower compared to the standard protocol (p = 0.03). The same pattern was observed with regard to SNR (Table 3).

Except where indicated, data represent numbers of patients and numbers in parentheses are percentages. * Data are given as mean ± standard deviation and range in parentheses.

site. In the other 15 patients, retrograde puncture of the common femoral artery was followed by a cross-over maneuver, providing images of the iliac arteries as well. In the remaining nine patients, the vasculature of both legs was completely depicted by DSA. Altogether, 888 segments were depicted by DSA and could be included in the analysis. No adverse events were recorded during the study.

3.2. Radiation exposure When compared to the low-dose ISP, application of the BMI-adjusted ultra-low-dose protocol facilitated a decrease of CTDIvol from a 152

European Journal of Radiology 93 (2017) 149–156

M.M. Schreiner et al.

Fig. 3. Head to head comparison of the image quality of the BMI groups. Each panel shows an axial CTA image at the level of the right common iliac artery, the left panel of an 85-year-old female patient with a BMI of 18.9, the right panel of a 75-year-old male patient with a BMI of 28.4. Despite the difference in BMI, the vascular enhancement as well as the image noise was similar, resulting in no significant difference in contrast- and signal-to-noise ratio.

respectively, were observed. In group 2 (25 < BMI ≤ 30), a mean CTDIvol and DLP of 2.55 ± 0.39 mGy and 337 ± 59 mGy x cm, respectively, were recorded (Fig. 4).

Table 3 Intra-individual (n = 15) comparison of the signal-to-noise-ratios (SNR) and contrast-tonoise-ratios (CNR) on axial images between the BMI-adjusted ultra-low-dose protocol and the institutional standard protocol (ISP).

CNR

SNR

Vessel regions

Ultra-low-dose protocol

ISP

p-value

3.3. Diagnostic performance

Iliac bifurcation (n = 30) Femoral bifurcation (n = 29) Tibial peroneal trunk (n = 26) All regions (n = 85)

8.17 ± 2.04

9.66 ± 3.22

0.03

21.03 ± 7.38

21.45 ± 7.48

0.80

26.60 ± 10.08

27.27 ± 10.35

0.79

18.20 ± 10.50

19.07 ± 10.42

0.37

Iliac bifurcation (n = 30) Femoral bifurcation (n = 29) Tibial peroneal trunk (n = 26) All regions (n = 85)

9.32 ± 2.24

11.12 ± 3.39

0.01

23.69 ± 8.27

24.63 ± 8.02

0.60

29.93 ± 11.31

30.77 ± 11.15

0.75

20.52 ± 11.76

21.74 ± 11.44

0.23

Of 888 segments visualized by DSA, 869 (97.9%) and 857 (96.5%) segments could be assessed with CTA by readers 1 and 2, respectively. Reader 1 rated 19 (2.1%) segments non-diagnostic (nine due to prosthesis-related artifacts and 10 due to severe circumferential calcifications). Reader 2 rated 31 (3.5%) segments non-diagnostic in CTA (10 for low contrast, eight due to prosthesis-related artifacts, one due to stent-related artifacts, and 12 due to severe circumferential calcifications). Based on the CT evaluation of Reader 1, overall sensitivity, specificity, and accuracy for the detection of hemodynamically relevant stenoses ( > 70%) were 92%, 96%, and 94%, respectively. For Reader 2, overall sensitivity, specificity and accuracy were all 93% as depicted in Table 4. Looking at the vessel territories in detail, sensitivity, specificity, accuracy, PPV, and NPV of both readers were higher in the infra-popliteal arteries compared to the femoro-popliteal arteries. For the iliac arteries, the results were less consistent and differed more between readers as well as in comparison to other territories (Table 4). Interreader agreement was good, as demonstrated by a kappa value of 0.77.

All data are given as mean ± standard deviation.

mean of 4.18 ± 0.62 mGy (range, 2.95–5.35) to a mean of 1.97 ± 0.55 mGy (range: 1.50–3.36) per scan, corresponding to a decrease of 53% (p < 0.001). Correspondingly, the mean DLP value dropped significantly (p < 0.001) from 544 ± 83 mGy x cm (range, 435–733) to 256 ± 81 mGy x cm (range: 178–444). Furthermore, significant differences between BMI groups were observed for both CTDIvol (p < 0.001) and DLP (p < 0.001). In group 1 (BMI ≤ 25), a mean CTDIvol and DLP of 1.69 ± 0.27 mGy and 214 ± 38 mGy x cm,

4. Discussion The main finding of our study was that a CTA protocol that 153

European Journal of Radiology 93 (2017) 149–156

M.M. Schreiner et al.

Fig. 4. A 74-year-old male patient with a BMI of 25.2 was referred to interventional radiology due to PAD stage IIb. According to our study protocol, the patient underwent preinterventional CTA with 80 kV and a reference mAs of 150, resulting in a CTDIvol of 2.23 mGy and a DLP of 272 mGy x cm. (a) In the middle part of the superficial femoral artery, a stent with hemodynamically significant in-stent stenosis was detected (white arrowheads), followed by an occluded stent (black arrowheads). (b) Both findings were confirmed by intra-procedural DSA. (c) Revascularization was successful after PTA of the in-stent-stenosis and placement of an additional stent in the occluded segment. Sufficient outflow in the lower leg was provided by the anterior tibial and the peroneal artery, showing no hemodynamically significant stenosis on CTA (d) or DSA (e).

tube current, and iterative image reconstruction allows for a significant reduction in radiation dose in comparison to our institutional standard low-dose protocol. In direct comparison to the study by Iezzi et al. [29], who included patients with a BMI of up to 30 as well, the mean CTDIvol of our institutional standard low-dose protocol matched their low-dose protocol, whereas the BMI-adjusted ultra-low-dose protocol facilitated an additional reduction in radiation dose of 65%. And for the sub-group with a BMI smaller than 25, which was similar to the cohort of Duan et al. [30], the radiation exposure was 55% lower. Furthermore, BMIadjusted modulation of the tube current allowed for a favorable personalization of the radiation dose, while maintaining image quality, as demonstrated by similar CNR and SNR in both BMI groups. Despite these reductions in radiation exposure, a promising diagnostic performance was still achieved. In previous studies, the diagnostic performance of CTA in below-the-knee arteries was reportedly lower than that for the aorto-iliac and femoro-popliteal arteries [11]. Different results, however, were observed in our study. For both readers, most of the assessed parameters of diagnostic performance were similar for the infra-popliteal arteries compared to the femoropopliteal arteries. We attribute this finding mainly to the increase in spatial resolution provided by the most recent CT scanners, thus improving the delineation of small vessels. Regarding the iliac arteries, the interpretation of our results is more difficult, as the limited number of segments in this region reduces the power of our findings. In addition, the percentage of diseased vessel segments in this vessel territory (12%) was considerably lower compared to the other vessel territories in our study (22% and 42% in the femoro-popliteal and infrapopliteal arteries, respectively), which particularly affected the positive predictive value.

incorporates low kV settings with a personalized (BMI-adjusted) reference amplitude for tube current modulation and iterative reconstruction, facilitates a significant reduction of radiation exposure, while maintaining good image quality and high diagnostic accuracy in the assessment of PAD. The impact of a lower tube voltage on diagnostic accuracy has been evaluated in a few studies [29,30]. Iezzi et al. [29] included patients with a BMI of up to 30 and compared multiple tube voltage presets of a 64-row CT scanner in a prospective, randomized trial. These authors demonstrated that reducing the tube voltage from 120 kV to 80 kV allowed for a substantial reduction of CTDIvol from 12.96 mGy to 5.69 mGy (61%), while maintaining diagnostic performance. Using a tube voltage of 70 kV, Duan et al. [30] reported an even lower mean CTDIvol of 3.71 mGy and an excellent diagnostic performance. However, only patients with a BMI < 25 were included. Unfortunately, this study cohort does not reflect the high prevalence of obese PAD patients in the Western world, and it is reasonable to conclude, that a tube voltage of 70 kV might not yield sufficient image quality in patients with a BMI greater than 25, particularly in the pelvic area. In addition to lower tube voltage settings, a decreased tube current has the potential to reduce radiation exposure as well, while maintaining diagnostic accuracy, as reported by Fraioli [22]. In our study, patients with a BMI of up to 30 were included and BMI was used to adapt the reference amplitude for tube current modulation accordingly. Even though other patient characteristics might be more accurate for this adaptation (e.g., sagittal abdominal diameter), we decided to adapt for the clinically easily assessable and applicable BMI. Our results show that this combination of low tube voltage, BMI-adjusted modulation of 154

European Journal of Radiology 93 (2017) 149–156

M.M. Schreiner et al.

diagnostic performance were in the range of most published studies [43], whereas the radiation exposure was lower compared to previously published studies [29,30]. Furthermore, regarding image quality and radiation exposure we were able to perform a statistically favorable intraindividual comparison with a subgroup of patients, who had previously underwent CTA with our institutional standard low dose protocol. In conclusion, the application of a low-dose protocol that incorporates personalized BMI-adjusted tube-current selection, a low tube voltage of 80 kV, and iterative image reconstruction seems promising. Personalization of the imaging protocol by adjusting the reference tube current according to BMI is beneficial for achieving the optimal tradeoff between radiation exposure and image quality.

Table 4 Diagnostic Performance of BMI-adjusted ultra-low-dose CT angiography with digital subtraction angiography as the reference standard according to vascular territory. Reader 1

Reader 2

Iliac arteries Sensitivity Specificity Accuracy PPV NPV

100 (12/12) 96 (78/81) 97 (90/93) 80 (12/15) 100 (78/78)

82 91 90 56 97

(9/11) (74/81) (83/92) (9/16) (74/76)

Femoro-popliteal arteries Sensitivity Specificity Accuracy PPV NPV

87 94 92 79 96

(81/93) (302/323) (383/416) (81/102) (302/314)

86 93 91 77 96

(80/93) (299/323) (379/416) (80/104) (299/312)

Infra-popliteal arteries Sensitivity Specificity Accuracy PPV NPV

94 99 97 98 96

(142/151) (206/209) (348/360) (142/145) (206/215)

99 93 95 91 99

(143/145) (190/204) (333/349) (143/157) (190/192)

Overall Sensitivity Specificity Accuracy PPV NPV

92 96 94 90 97

(235/256) (586/613) (821/869) (235/262) (586/607)

93 93 93 84 97

(232/249) (563/608) (795/857) (232/277) (563/580)

Conflict of interest None Funding This work was supported by the Austrian Science Fund (FWF) [grant number TRP 67-N23]. IRB statement This study was approved by the local institutional review board and was conducted according to the Declaration of Helsinki, including current revisions. Written, informed consent was obtained from all patients prior to recruitment.

Data are given as percentages; numerators and denominators are displayed in parentheses; PPV = positive predictive value, NPV = negative predictive value.

However, it is worth noting that adjustment of the tube current for BMI facilitated a similar image quality in the iliac arteries in both BMI groups. This confirms that the patient’s BMI is a good parameter to personalize the tube current in order to maintain constant image quality while keeping radiation exposure as low as reasonably achievable. Similar to CTA, MR angiography (MRA) has significantly improved over the last decades and has become another alternative for the noninvasive assessment of the peripheral arteries [31]. Its key advantage over CTA is without doubt that it lacks the biggest drawback of CTA − ionizing radiation. Although new acquisition sequences are currently being developed to facilitate MRA with good diagnostic accuracy, but without the necessity to inject MR contrast media containing gadolinium chelates [32], contrast-enhanced moving-bed MR angiography is still most often performed in clinical routine [33]. This is of particular importance as certain types of MR contrast media have not only been linked to nephrogenic systemic fibrosis [34] but are now also discussed with regard to gadolinium depositions in the brain [35]. Furthermore, MRA is contraindicated in patients with certain implanted medical products (e.g. pacemakers, CSF shunts) [36]. Additional advantages of CTA over MRA include better depiction of in-stent-stenosis [37], costsavings [38] as well as faster acquisition, which renders CTA the preferred modality of most patients [39]. Although vessel wall calcifications can affect the diagnostic accuracy of peripheral CTA [40], the knowledge about of these calcifications is important for the planning of subsequent endovascular or surgical treatment [41]. While standard MRA sequences lack the ability to depict these calcifications, a promising new MRA sequence was recently developed to overcome this limitation [42]. Our study had several limitations. First, not all segments were visualized by DSA due to ethical considerations. However, this was considered at the a priori power analysis. Second, there is no prospective, randomized comparison with a control group that underwent CTA with a standard 80 kV protocol and subsequent DSA. Thus, possible differences in diagnostic performance and radiation exposure could not be elucidated. However, the statistical parameters of

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ejrad.2017.06.002. References [1] W.T. Meijer, A.W. Hoes, D. Rutgers, M.L. Bots, A. Hofman, D.E. Grobbee, Peripheral arterial disease in the elderly: The Rotterdam Study, Arterioscler. Thromb. Vasc. Biol. 18 (2) (1998) 185–192. [2] E. Selvin, T.P. Erlinger, Prevalence of and risk factors for peripheral arterial disease in the United States: results from the National Health and Nutrition Examination Survey, 1999–2000, Circulation 110 (6) (2004) 738–743. [3] L. Norgren, W.R. Hiatt, J.A. Dormandy, M.R. Nehler, K.A. Harris, F.G. Fowkes, Inter-society consensus for the management of peripheral arterial disease (TASC II), J. Vasc. Surg. 45 (Suppl S) (2007) S5–67. [4] N. Young, K.K. Chi, J. Ajaka, L. McKay, D. O'Neill, K.P. Wong, Complications with outpatient angiography and interventional procedures, Cardiovasc. Intervent. Radiol. 25 (2) (2002) 123–126. [5] M. Romano, P.P. Mainenti, M. Imbriaco, et al., Multidetector row CT angiography of the abdominal aorta and lower extremities in patients with peripheral arterial occlusive disease: diagnostic accuracy and interobserver agreement, Eur. J. Radiol. 50 (3) (2004) 303–308. [6] R. Schernthaner, A. Stadler, F. Lomoschitz, et al., Multidetector CT angiography in the assessment of peripheral arterial occlusive disease: accuracy in detecting the severity, number, and length of stenoses, Eur. Radiol. 18 (4) (2008) 665–671. [7] X.M. Li, Y.H. Li, J.M. Tian, et al., Evaluation of peripheral artery stent with 64-slice multi-detector row CT angiography: prospective comparison with digital subtraction angiography, Eur. J. Radiol. 75 (1) (2010) 98–103. [8] M.C. Kock, M.E. Adriaensen, P.M. Pattynama, et al., DSA versus multi-detector row CT angiography in peripheral arterial disease: randomized controlled trial, Radiology 237 (2) (2005) 727–737. [9] R. Schernthaner, D. Fleischmann, F. Lomoschitz, A. Stadler, J. Lammer, C. Loewe, Effect of MDCT angiographic findings on the management of intermittent claudication, AJR Am. J. Roentgenol. 189 (5) (2007) 1215–1222. [10] R. Schernthaner, D. Fleischmann, A. Stadler, M. Schernthaner, J. Lammer, C. Loewe, Value of MDCT angiography in developing treatment strategies for critical limb ischemia, AJR Am. J. Roentgenol. 192 (5) (2009) 1416–1424. [11] R. Ouwendijk, M.C. Kock, L.C. van Dijk, M.R. van Sambeek, T. Stijnen, M.G. Hunink, Vessel wall calcifications at multi-detector row CT angiography in patients with peripheral arterial disease: effect on clinical utility and clinical predictors, Radiology 241 (2) (2006) 603–608. [12] G.D. Rubin, Data explosion: the challenge of multidetector-row CT, Eur. J. Radiol. 36 (2) (2000) 74–80.

155

European Journal of Radiology 93 (2017) 149–156

M.M. Schreiner et al. [13] A. Dohan, R. Dautry, Y. Guerrache, et al., Three-dimensional MDCT angiography of splanchnic arteries: pearls and pitfalls, Diagn. Interv. Imag. 96 (2) (2015) 187–200. [14] T. Kau, W. Eicher, C. Reiterer, et al., Dual-energy CT angiography in peripheral arterial occlusive disease-accuracy of maximum intensity projections in clinical routine and subgroup analysis, Eur. Radiol. 21 (8) (2011) 1677–1686. [15] D.N. Tran, M. Straka, J.E. Roos, S. Napel, D. Fleischmann, Dual-energy CT discrimination of iodine and calcium: experimental results and implications for lower extremity CT angiography, Acad. Radiol. 16 (2) (2009) 160–171. [16] J.E. Roos, D. Fleischmann, A. Koechl, et al., Multipath curved planar reformation of the peripheral arterial tree in CT angiography, Radiology 244 (1) (2007) 281–290. [17] D.J. Brenner, E.J. Hall, Computed tomography–an increasing source of radiation exposure, N. Engl. J. Med. 357 (22) (2007) 2277–2284. [18] E.J. Hall, D.J. Brenner, Cancer risks from diagnostic radiology: the impact of new epidemiological data, Br. J. Radiol. 85 (1020) (2012) e1316–7. [19] D.P. Frush, Strategies of dose reduction, Pediatr. Radiol. 32 (4) (2002) 293–297. [20] M.J. Willemink, T. Leiner, P.A. de Jong, et al., Iterative reconstruction techniques for computed tomography part 2: initial results in dose reduction and image quality, Eur. Radiol. 23 (6) (2013) 1632–1642. [21] B. Wintersperger, T. Jakobs, P. Herzog, et al., Aorto-iliac multidetector-row CT angiography with low kV settings: improved vessel enhancement and simultaneous reduction of radiation dose, Eur. Radiol. 15 (2) (2005) 334–341. [22] F. Fraioli, C. Catalano, A. Napoli, et al., Low-dose multidetector-row CT angiography of the infra-renal aorta and lower extremity vessels: image quality and diagnostic accuracy in comparison with standard DSA, Eur. Radiol. 16 (1) (2006) 137–146. [23] W. Huda, E.M. Scalzetti, G. Levin, Technique factors and image quality as functions of patient weight at abdominal CT, Radiology 217 (2) (2000) 430–435. [24] K.T. Bae, H.Q. Tran, J.P. Heiken, Uniform vascular contrast enhancement and reduced contrast medium volume achieved by using exponentially decelerated contrast material injection method, Radiology 231 (3) (2004) 732–736. [25] D. Beitzke, R. Nolz, S. Unterhumer, et al., Low-dose high-pitch CT angiography of the supraaortic arteries using sinogram-affirmed iterative reconstruction, PLoS One 9 (6) (2014) e99832. [26] A. Winklehner, C. Karlo, G. Puippe, et al., Raw data-based iterative reconstruction in body CTA: evaluation of radiation dose saving potential, Eur. Radiol. 21 (12) (2011) 2521–2526. [27] P.B. Noel, B. Renger, M. Fiebich, et al., Does iterative reconstruction lower CT radiation dose: evaluation of 15,000 examinations, PLoS One 8 (11) (2013) e81141. [28] J.R. Landis, G.G. Koch, The measurement of observer agreement for categorical data, Biometrics 33 (1) (1977) 159–174. [29] R. Iezzi, M. Santoro, R. Marano, et al., Low-dose multidetector CT angiography in the evaluation of infrarenal aorta and peripheral arterial occlusive disease, Radiology 263 (1) (2012) 287–298. [30] Y. Duan, X. Wang, X. Yang, D. Wu, Z. Cheng, L. Wu, Diagnostic efficiency of lowdose CT angiography compared with conventional angiography in peripheral arterial occlusions, AJR Am. J. Roentgenol. 201 (6) (2013) W906–14. [31] J. Menke, J. Larsen, Meta-analysis: accuracy of contrast-enhanced magnetic

[32]

[33]

[34]

[35] [36]

[37]

[38]

[39]

[40]

[41] [42]

[43]

156

resonance angiography for assessing steno-occlusions in peripheral arterial disease, Ann. Intern. Med. 153 (5) (2010) 325–334. P.A. Hodnett, E.V. Ward, A.H. Davarpanah, et al., Peripheral arterial disease in a symptomatic diabetic population: prospective comparison of rapid unenhanced MR angiography (MRA) with contrast-enhanced MRA, AJR Am. J. Roentgenol. 197 (6) (2011) 1466–1473. C. Loewe, M. Schoder, T. Rand, et al., Peripheral vascular occlusive disease: evaluation with contrast-enhanced moving-bed MR angiography versus digital subtraction angiography in 106 patients, AJR Am. J. Roentgenol. 179 (4) (2002) 1013–1021. T. Grobner, Gadolinium–a specific trigger for the development of nephrogenic fibrosing dermopathy and nephrogenic systemic fibrosis? Nephrol. Dial. Transplant. 21 (4) (2006) 1104–1108. R.J. McDonald, J.S. McDonald, D.F. Kallmes, et al., Intracranial gadolinium deposition after contrast-enhanced MR imaging, Radiology 275 (3) (2015) 772–782. G.N. Levine, A.S. Gomes, A.E. Arai, et al., Safety of magnetic resonance imaging in patients with cardiovascular devices: an American Heart Association scientific statement from the Committee on Diagnostic and Interventional Cardiac Catheterization, Council on Clinical Cardiology, and the Council on Cardiovascular Radiology and Intervention: endorsed by the American College of Cardiology Foundation, the North American Society for Cardiac Imaging, and the Society for Cardiovascular Magnetic Resonance, Circulation 116 (24) (2007) 2878–2891. M.B. Schernthaner, G. Edelhauser, D. Berzaczy, et al., Perceptibility and quantification of in-stent stenosis with six peripheral arterial stent types in vitro: comparison of 16-MDCT angiography, 64-MDCT angiography, and MR angiography, AJR Am. J. Roentgenol. 194 (5) (2010) 1346–1351. R. Ouwendijk, M. de Vries, P.M. Pattynama, et al., Imaging peripheral arterial disease: a randomized controlled trial comparing contrast-enhanced MR angiography and multi-Detector row CT angiography, Radiology 236 (3) (2005) 1094–1103. J.K. Willmann, S. Wildermuth, T. Pfammatter, et al., Aortoiliac and renal arteries: prospective intraindividual comparison of contrast-enhanced three-dimensional MR angiography and multi-detector row CT angiography, Radiology 226 (3) (2003) 798–811. C. Catalano, F. Fraioli, A. Laghi, et al., Infrarenal aortic and lower-extremity arterial disease: diagnostic performance of multi-detector row CT angiography, Radiology 231 (2) (2004) 555–563. E. Ascer, F.J. Veith, S.A. Flores, Infrapopliteal bypasses to heavily calcified rock-like arteries. Management and results, Am. J. Surg. 152 (2) (1986) 220–223. M.P. Ferreira Botelho, I. Koktzoglou, J.D. Collins, et al., MR imaging of iliofemoral peripheral vascular calcifications using proton density-weighted, in-phase threedimensional stack-of-stars gradient echo, Magn. Reson. Med. 77 (6) (2017) 2146–2152. R. Met, S. Bipat, D.A. Legemate, J.A. Reekers, M.J. Koelemay, Diagnostic performance of computed tomography angiography in peripheral arterial disease: a systematic review and meta-analysis, JAMA 301 (4) (2009) 415–424.