Article Text
PDF
Abstract
Background High-frequency optical coherence tomography (HF-OCT) is an intravascular imaging method that allows for volumetric imaging of flow diverters in vivo.
Objective To examine the hypothesis that a threshold for both volume and area of communicating malapposition can be predictive of early aneurysm occlusion.
Methods Fifty-two rabbits underwent elastase aneurysm formation, followed by treatment with a flow diverter. At the time of implant, HF-OCT was acquired to study the rate and degree of communicating malapposition. Treated aneurysms were allowed to heal for either 90 or 180 days and euthanized following catheter angiography. Healing was dichotomized into aneurysm remnant or neck remnant/complete occlusion. Communicating malapposition was measured by HF-OCT using a semi-automatic algorithm able to detect any points where the flow diverter was more than 50 µm from the vessel wall. This was then summed across image slices to either a volume or area. Finally, a subsampled population was used to train a statistical classifier for the larger dataset.
Results No difference in occlusion rate was found between device type or follow-up time (p=0.28 and p=0.67, respectively). Both volume and area of malapposition were significantly lower in aneurysms with a good outcome (p<0.001, both). From the statistical model, a volume of less than 0.56 mm3 or a normalized area less than 0.69 as quantified by HF-OCT was predictive of occlusion (p<0.001, each).
Conclusions HF-OCT allows for measurements of both volume and area of malapposition and, from these measurements, an accurate prediction for early aneurysm occlusion can be made.
- Flow Diverter
- Aneurysm
- Device
- Intervention
- Technology
Data availability statement
Data are available upon reasonable request. Data are available by contacting the corresponding author.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Prior preclinical studies have provided strong evidence that flow diverter malapposition at the level of the aneurysm neck is a cause of delayed healing.
WHAT THIS STUDY ADDS
Although malapposition can be detected by intravascular imaging with high-frequency optical coherence tomography (HF-OCT), quantitative thresholds that predict early aneurysm occlusion have not been established. This study shows that both volume and area measurements of the extent of malapposition can predict early aneurysm occlusion with high sensitivity and specificity.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
With the introduction of HF-OCT to patients imminent, this preclinical research provides evidence to formulate hypotheses for pending clinical studies. Ultimately, HF-OCT measurements in patients treated with flow diversion may provide actionable information to improve device apposition to ensure reliable early aneurysm exclusion from the circulation.
Introduction
Flow diversion has shown high levels of treatment success at long-term follow-up of aneurysm healing.1 However, short-term healing rates tend to be much lower, during which aneurysms remain vulnerable for rupture.2 With the development of neurovascular high-frequency optical coherence tomography (HF-OCT),3 4 the ability to image and measure vessel wall apposition while maintaining high spatial accuracy even in areas of elevated tortuosity is now achievable. Recently, many studies have examined the ability to predict intracranial aneurysm occlusion after flow diverter treatment: computational fluid dynamics (CFD),5 6 artificial intelligence algorithms,7–9 and angiographic metrics10–12; however, all of these are based on fluoroscopic imaging, which has a resolution one order of magnitude below that of HF-OCT.
Although malapposition occurs at any point along the stent, and can be associated with various complications, this work focuses on communicating malapposition (CM)13 and the ability to assess a dynamic range of associated measurements. After flow diverter placement, HF-OCT imaging allows for the quantification of CM area and volume—that is, the area and volume between the flow diverter and the vessel wall. In this study, we hypothesized that these quantitative measurements of CM are predictors of poor aneurysm occlusion rates at 90 and 180 days. We further hypothesized that these measurements allow for a statistical model to be created that can predict the failure of aneurysm healing.
Methods
All animal research was approved by our Institutional Animal Care and Use Committee. Fifty-two New Zealand White rabbits (either sex, weight ranges 2.8–4.0 kg) underwent elastase-induced aneurysm creation of the right common carotid artery, as described elsewhere.14 The nascent aneurysms were allowed to form over a minimum of 21 days, after which the rabbits were randomized for treatment. Treatment was divided into five different device types (variations of surface coatings to the Evolve flow diverter; Stryker Neurovascular, Fremont, California, USA) that were blinded to the investigative team, and two implant durations (90 or 180 days). Five days before implant, all animals were given 10 mg/kg aspirin and clopidogrel, which continued for at least 30 days post-implant.
All procedures were performed under general anesthesia and with strict aseptic technique. Prior to all surgical or imaging procedures, the animals were pre-anesthetized by a subcuticular injection of glycopyrrolate (0.01 mg/kg) and given an intramuscular dose of buprenorphine SR (0.15 mg/kg) for pain management. Anesthesia was induced by an intramuscular injection of ketamine (35 mg/kg) and xylazine (5 mg/kg) and maintained with mechanical ventilation of 1–3% isoflurane. The physiological status of the animal was assessed using continuous monitoring of respiration rate, heart rate, oxygen saturation level, end-tidal CO2 level, and temperature.
Once anesthetized, the right femoral artery was exposed through a 2 cm incision. A 6-French introducer sheath was inserted over a guidewire and through an arteriotomy of the femoral artery. A bolus of 100 U/kg heparin was administered. Following the heparin administration, a 6 F Guider Softip (Boston Scientific, Marlborough, Massachusetts, USA) was advanced to the ostium of the innominate, and digital subtraction angiography (DSA) was performed to measure the dimensions of the brachiocephalic and subclavian arteries to properly size the flow diverter. A flow diverter implant was performed by an experienced interventionalist (VA) who was blinded to the assigned animal group. After the flow diverter was successfully deployed, HF-OCT was performed to assess the apposition of the device. To adequately clear the blood during the HF-OCT pullback, contrast media was injected at a rate of 3.5 mL/s over a period of 5 s. Once HF-OCT imaging was completed, the sheath was removed, and the animal was recovered. Depending on the group assignment, the rabbits were returned for imaging at either 90 or 180 days post-implant and occlusion was assessed on a 5-point angiographic scale, as previously described.15
To leverage the increased resolution of HF-OCT, a semi-automated image processing pipeline was developed. Trained users were required to identify the lumen of the vessel and the struts of the flow diverter. Thereafter, an automated process was created to identify the aneurysm, and any malapposition between the flow diverter and lumen at the level of the aneurysm neck (figure 1). Based on the thickness of the stent struts,16 a threshold of 50 µm (approximately 1.5 times the strut thickness) was used to define malapposition. This automated process was repeated for all slices in the HF-OCT pullback that included CM, and the measurements were integrated to turn the linear measurement (figure 1, gray) into an area, and the area (figure 1, green) into a volume. The same process was used to measure the area of the aneurysm neck (figure 1, blue). Since the primary outcome of this study was the occlusion rates of aneurysms, only CM13 was measured. This quantitative measurement resulted in three parameters of interest: (1) area of the aneurysm neck, (2) area of the annular CM, and (3) volume of malapposition between the flow diverter and the vessel wall. The area of the CM was normalized to the aneurysm neck area to control for differences in aneurysm neck sizes and compared against the five progressive healing scores. The segmentation of HF-OCT images and assessment of DSA was done by two independent reviewers. The retaliative strength of both methods was compared using a receiver operating characteristic (ROC) analysis.17
High-frequency optical coherence tomography slice of a malapposed device. The gray arc depicts the area of the malapposition, the green area between the flow diverter and vessel wall is the volume of malapposition, and the blue arc is the aneurysm neck. *Aneurysm sac.
Finally, the ability of HF-OCT to predict whether an aneurysm is occluded was assessed based on a statistical model. First, the full dataset was subsampled such that approximately one-third of the animals, chosen randomly, were used to train the model. To do this, the average and standard deviation (SD) for both the malapposed volume and the normalized CM area was calculated from the subsampled populations. This was then used to classify the full population as either poor outcome (angiographic scores 0–2) or good outcome (angiographic scores 3–4). Sensitivity and specificity of the classifier were calculated, and the process was repeated 10 times to ensure absence of bias based on the random subsample.
All statistical calculations were done in either R4.2.0 (the R foundation) or MATLAB 2021a (MathWorks, Natick, Massachusetts, USA), with a p value of less than 0.05 considered to be significant, and a Shapiro-Wilk normality test was applied. When an analysis of variance (ANOVA) was used, Tukey post hoc analysis was conducted to determine which group and factors were statistically significant.
Results
Of 52 rabbits, 43 (83%) had both adequate HF-OCT and successful device placement to be considered for evaluation. Inadequate HF-OCT (n=6) because of failure to clear blood during image acquisition; and due to the small size of rabbits, repeat contrast injections were permissible. In three cases, the device did not cover the entire aneurysm neck and hence were excluded from the analysis. To group all the animals together, the differences between the 90- and 180-day occlusion rates were compared, and no differences were found (p=0.67). This allowed for combination of the two timepoints when assessing the effects of CM. Similarly, no differences were seen in the occlusion rates between the five different device categories (p=0.12, Χ2 test), or in the dichotomized occlusion rate between devices (p=0.28, table 1) and thus were combined for the analysis. Occlusion rates of both CM volume and normalized CM area were found to be significantly different across the groups (p<0.001 both, ANOVA), and in a subgroup analysis (figure 2), the rate of normalized CM area was significantly lower in both the aneurysms scored as 3 and 4 (p<0.01, Tukey post hoc). The same scenario was observed for the volume of CM (p<0.01) (figure 3).
The top row depicts a well-apposed device: (A) Pre-implant digital subtraction angiography (DSA), (B) post-implant high-frequency optical coherence tomography (HF-OCT) showing only slight malapposition, and (C) 90-day follow-up DSA revealing an occluded aneurysm (score=4). The bottom row shows an example of a device with a large malapposition on HF-OCT: (D) A similar pre-implant DSA, (E) HF-OCT slice showing the malapposition, and (F) the aneurysm still filling at the 90-day follow-up DSA (score=2).
Malapposition volume (A) and normalized area (B) arranged by final angiographic occlusion score. For both volume and area, a score of 3–4 (good outcome) was significantly lower than that of a score of 0–1 (very poor outcome). **p<0.01.
Baseline characteristics of implanted aneurysm and overall occlusion rate, stratified by device type
Beyond the differences in CM volume and normalized area assessed by healing scores, these HF-OCT parameters were also used to assess the prediction of healing. Calculating the area under the curve (AUC) using an ROC analysis (figure 4) showed a high AUC for both the CM volume and normalized area (0.92 and 0.95, respectively); furthermore, no significant differences were observed between the two analysis methods (95% CI 0.85 to 1.0 vs 0.88 to 1.0, CM volume and normalized area).
Receiver operating characteristic analysis for both the volume of malapposition (blue) and normalized area (dashed red). Both show a very high area under the curve (0.92 and 0.95, respectively), indicating that both methods can accurately predict if an aneurysm will occlude.
On validation of significant differences at each healing score and the high AUC, parameters measured with HF-OCT were used to predict if an aneurysm would heal. Rabbits were divided into two groups based on the DSA occlusion score: a score of 3–4 was considered a good outcome (n=21) and a score of 0–2 a poor outcome (n=22). Randomly subsampled populations were drawn from both the good outcome group (n=7, 33%) and the poor outcome group (n=8, 36%) and used to calculate the thresholds between good and poor outcome for both volume and normalized area of CM. For the volume of CM, a threshold value less than 0.56 mm3 was found to be predictive of a good outcome with a sensitivity of 0.83 and specificity of 0.87; however, the normalized area had both a better sensitivity and specificity (0.90 and 0.92, respectively) with a threshold value of less than 0.69 being predictive of a good outcome. Both methods showed a significant ability to predict the outcome based on the semi-automated segmentation algorithm (p<0.001 for each), with the normalized area showing slightly stronger results.
Discussion
Previous works on malapposition assessment have been limited to either histological studies18 ,19 or simple binary scores owing to a lack of true 3D imaging volumes.13 HF-OCT is a powerful technique allowing for a densely sampled 3D imaging volume that reduces partial volume errors. By using a high-resolution volumetric imaging technique, we can investigate the effect of small areas of malapposition in aneurysm healing and devise a thresholding method to determine the impact of CM on aneurysm healing. Herein, we showed that a volume of less than 0.56 mm3 of malapposition was small enough to not affect the ability of the flow diverter to heal. This was also true for an area of less than 69% of the neck of the aneurysm. Both analysis methods showed significant predictive ability.
Imaging data have been used to predict the healing of aneurysms. The simplest method is to look at post-implant DSA and contrast stasis; however, this method is dependent on the orientation of the patient and is often unreliable.10 20 21 A more sophisticated approach is to use an optical flow tracker on contrast transport in the aneurysm,22 and thereby calculating the mean aneurysm flow amplitude change between pre- and post-diverter implant, which has been shown to be predictive of aneurysm healing.23 It is highly probable that the mean aneurysm flow amplitude is a surrogate measure of device malapposition, which HF-OCT can image directly. Another approach applies CFD to 3D data and looks for mean aneurysm velocity and flow rate,5 achieving a sensitivity and specificity of 80% and 75%, respectively, which compares well with other imaging-based aneurysm occlusion prediction methods.7 Here, we have shown both an increased sensitivity and specificity (90% and 92%, respectively), most probably owing to the higher spatial resolution of HF-OCT with data available in real-time and without using hemodynamic information. This opens the possibility of potentially applying CFD in combination with HF-OCT to further increase the prediction accuracy. CFD is highly influenced by both the flow conditions and the geometric model of the flow diverter; imaging that can identify the patient-specific conformation of the flow diverter, and not an idealized model, will potentially improve the accuracy of such simulations.
Recent literature on the cause of aneurysmal healing after flow diversion has pointed to the importance of the endothelial cells within the parent artery.24 This might help to explain why CM is of such importance to the rate of healing, as one of the primary functions of a flow diverter is to create an endoluminal support scaffold for the nascent vessel wall to grow and permanently exclude the aneurysm from the circulation. When there is CM present, the flow diverter will see a higher level of flow across the struts, which could slow down the rate of new endothelial growth.25 HF-OCT allows for imaging the degree of CM when the flow diverter is implanted, and thus opens up the possibility of treating any malapposition that is seen. This is further reinforced by the ROC analysis, where it might be possible to investigate the two different components of healing. The volume of malapposition is most probably related to the ability of the flow diverter to reduce the momentum transfer into the aneurysm sac, whereas the area of malapposition is more probably related to the degree and rate of endothelialization, by increasing the distance that new tissue growth must cover before the new blood vessel is grown. Undoubtedly, there is correlation between these two different measurements and healing paradigms.
In this study, an artificial aneurysm is created by elastinolysis of the right common carotid artery of rabbits, which is different from the pathological conditions that cause the formation of an aneurysm in humans. Although the pathophysiology of this aneurysm model differs significantly from the disease in humans, once formed the rabbit elastase aneurysm has been shown to replicate the wall biology of the human intracranial aneurysms quite well.26 27 Unlike human aneurysms, rabbit elastase aneurysms show a high level of uniformity that make it an excellent research tool by reducing confounding variables. This rabbit aneurysm model has been used quite extensively in device testing and become the most common for preclinical testing of treatment devices, such as flow diverters.28 One major limitation is that these rabbit aneurysms do not spontaneously rupture, or grow,29 without modification.30 However, the results presented here are related to the question of how to accelerate the healing process and avoid delayed rupture, thus a lack of disease progression should not affect the conclusions.
The current study does have some limitations. Although no differences were seen in healing rates between the devices, it is possible that there was an unseen effect. HF-OCT measurements were taken after implant; however, the time between implant and HF-OCT was not controlled. It is foreseeable that the stent had not completely expanded, and a delayed HF-OCT would have better demonstrated the true final apposition. Taking all this into account, with such a high AUC (0.92 and 0.95) and sensitivity and specificity (90% and 92%, respectively), it is unlikely that any of these limitations would have significantly affected the outcome.
Conclusion
HF-OCT allows for the quantification of volume and area of flow diverter malapposition, both of which are strongly associated with device failure to lead to early aneurysm occlusion. A communicating malapposition volume smaller than 0.56 mm3 showed specificity and sensitivity of 0.87 and 0.83, respectively, to predict early aneurysm occlusion. Normalized area measurement of communicating malapposition, with a value less than 0.69, was an even stronger predictor of early aneurysm occlusion, showing a specificity and sensitivity of 0.92 of 0.90, respectively.
Supplemental material
Data availability statement
Data are available upon reasonable request. Data are available by contacting the corresponding author.
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Twitter @AjitSPuri1
Contributors RMK, MSS: responsible for data acquisition, data analysis, and statistical analysis. Drafted the manuscript. They are guarantors of the report. AP, CMR: responsible for data acquisition and analysis. VA, JMA, HGD, and MJG: responsible for planning, conception and design of the study, acquisition of data, and revising the manuscript. GJU and ASP: major contributions to high-frequency optical coherence tomography technology and data interpretation. Provided critical editing of the manuscript. All authors approved the final version of this manuscript to be published and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Funding This study was sponsored by Stryker Neurovascular and a Bits-2-Bytes grant from the Massachusetts Life Sciences Center. The content is solely the responsibility of the authors and does not necessarily represent the official views of the sponsors.
Competing interests ASP: consultant for Medtronic Neurovascular, Stryker Neurovascular, Balt, Q’Apel Medical, Cerenovus, Microvention, Imperative Care, Agile, Merit, CereVasc, and Arsenal Medical; research grants from NIH, Microvention, Cerenovus, Medtronic Neurovascular, and Stryker Neurovascular; holds stocks in InNeuroCo, Agile, Perfuze, Galaxy and NTI. JMA and HGD are employees of Stryker Neurovascular. GJU is an employee of Gentuity. MJG: has been a consultant on a fee-per-hour basis for Alembic LLC, Astrocyte Pharmaceuticals, BendIt Technologies, Cerenovus, Imperative Care, Jacob’s Institute, Medtronic Neurovascular, Mivi Neurosciences, phenox GMbH, Q’Apel, Route 92 Medical, Scientia, Stryker Neurovascular, Stryker Sustainability Solutions, Wallaby Medical; holds stock in Imperative Care, InNeuroCo, Galaxy Therapeutics and Neurogami; and has received research support from the National Institutes of Health (NIH), the United States – Israel Binational Science Foundation, Agile, Anaconda, ApicBio, Arsenal Medical, Axovant, Balt, Cerenovus, Ceretrieve, CereVasc LLC, Cook Medical, Galaxy Therapeutics, Gentuity, Gilbert Foundation, Imperative Care, InNeuroCo, Insera, Jacob’s Institute, Magneto, MicroBot, Microvention, Medtronic Neurovascular, MIVI Neurosciences, Naglreiter MDDO, Neurogami, Philips Healthcare, Progressive Medical, Pulse Medical, Rapid Medical, Route 92 Medical, Scientia, Stryker Neurovascular, Syntheon, ThrombX Medical, Wallaby Medical, the Wyss Institute and Xtract Medical. MJG is associate editor of Basic Science on the JNIS editorial board. MSS: has been a consultant on a fee-per-hour basis for Sanofi; and has research support from Sio Gene Therapies, Inozyme, the National Institutes of Health (NIH), Gilbert Foundation, and Massachusetts Life Science Center.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.