Intraoperative localization of cardiac conduction tissue regions using real-time fibre-optic confocal microscopy: first in human trial

Abstract

OBJECTIVES

The aim of this study was to evaluate the feasibility and safety of fibre-optic confocal microscopy (FCM) using fluorescein sodium dye for the intraoperative location of conduction tissue regions during paediatric heart surgery.

METHODS

The pilot study included 6 patients undergoing elective surgery for the closure of isolated secundum atrial septal defect aged 30 days to 21 years. FCM imaging was integrated within the normal intraoperative protocol for atrial septal defect repair. Fluorescein sodium dye was applied on the arrested heart. FCM images were acquired at the atrioventricular node region, sinus node region and right ventricle (RV). Total imaging time was limited to 3 min. Any adverse events related to the study were recorded and analysed. Subjects received standard postoperative care. Trained reviewers (n = 9) classified, de-identified and randomized FCM images (n = 60) recorded from the patients as presenting striated, reticulated or indistinguishable microstructures. The reliability of reviewer agreement was assessed using Fleiss’ kappa.

RESULTS

The FCM imaging instruments were integrated effectively into the cardiac surgery operating room. All adverse events found in the study were deemed expected and not related to FCM imaging. Reticulated myocardial microstructures were found during FCM imaging at atrioventricular node and sinus node regions, while striated microstructures were observed in RV. Reliability of agreement of reviewers classifying the FCM images was high (Fleiss’ kappa: 0.822).

CONCLUSIONS

FCM using fluorescein sodium dye was found to be safe for use during paediatric heart surgery. The study demonstrates the potential for FCM to be effective in identifying conduction tissue regions during congenital heart surgery.

Clinical trial registration number

NCT03189134.

INTRODUCTION

Congenital heart disease is the most common form of birth defects with ∼32 000 new cases occurring in the USA every year [1, 2] and 20 000 paediatric cardiac operations being performed annually to correct defects or palliate their effects [3]. One of the more common complications that can occur during congenital heart surgery is partial or complete heart block associated with interruption of cardiac conduction pathways [4]. The block can occur due to conduction tissue damage from trauma, suturing or other procedures. When permanent heart block occurs as a result of surgical repair, a permanent pacemaker is implanted allowing these patients to maintain atrioventricular synchrony. These patients, however, have an increased healthcare burden due to their dependence on the pacemaker implants for their entire lifetime. Because of implantation at a young age, the pacemaker implants require multiple revisions to accommodate patient size and subsequent growth and maturation [5, 6]. Over the patient’s lifetime, the associated lead and generator changes along with the costs and comorbidities place a continuing burden on the patients, their families and society [7, 8].

Difficulties identifying conduction tissue during cardiac surgery to repair congenital defects can contribute to conduction tissue damage and subsequent heart block [4]. Currently, congenital heart surgeons rely on anatomical landmarks to avoid damage to conduction tissue during corrective heart surgery [9]. These landmarks have been established over decades of pathology studies since the 1960s by pioneers in the field [10–12]. Recent high-resolution 3-dimensional representations of the cardiac conduction system utilizing contrast-enhanced micro-computed tomography [13] and whole-mount immunolabelling with volume imaging [14] represent new methods that when combined with historical anatomic knowledge enhance our understanding of the spatial distribution of conduction tissue and the cellular architecture of the nodal regions. Conduction pathways for many of the common congenital heart defects are well established with reliable roadmaps for congenital heart surgeons to follow during surgery [9, 12, 15]. However, the incidence rate of postoperative transient or complete heart block requiring temporary or permanent pacing has remained unchanged in the last 2 decades [4] indicating limitations of the current approach. Furthermore, in paediatric hearts with more complicated congenital defects like congenitally corrected transposition [16], straddling tricuspid valve and atrial isomerism/heterotaxy syndrome [17], intraoperative identification of key anatomical landmarks for conduction tissue localization can be challenging. Hence, a more accurate approach for intraoperative localization of conduction tissue during surgical intervention might significantly reduce the rate of conduction defects, especially in more complicated cases.

Based on immunohistochemical approaches for identifying cardiac conduction tissue [18, 19], we characterized tissue microstructural differences between atrial working myocardium and nodal tissue in rodent [20]. This study established that tissue microstructures in the regions of sinus (SN) and atrioventricular node (AVN) were reticulated while those in the working myocardium were striated. Furthermore, the endocardium/epicardium in these regions ranged in thickness from 2.8 to 10.3 µm. The subendocardial/subepicardial myocardium was therefore within the working distance of commercially available fibre-optic confocal microscopy (FCM) systems that are being utilized for in vivo biopsies in urogenital [21], gastrointestinal [22] and pulmonary [23] applications. We hypothesized that FCM can assist congenital heart surgeons in locating conduction tissue regions intraoperatively by enabling real-time imaging below the endocardium/epicardium to reveal distinguishable microstructures of cardiac conduction tissue. We translated this method in studies using rodent [24], ex vivo human [20] and ovine tissues [25]. We also assessed the method using expert human analysis [26].

Here, we report the first in human trial using the FCM system for locating conduction tissue regions. We utilized an off-label application of a US Food and Drug Administration (FDA)-approved imaging system and fluorescent dye under an investigational new drug exemption from the FDA. The primary objective of this study was to assure the safety of using diluted fluorescein sodium in conjunction with the FCM system during cardiac surgery. The secondary objective was to evaluate the logistical and technological feasibility of using the FCM imaging modality in the cardiothoracic operating room (OR). Our final objective was to assess the reliability of reviewers in classifying FCM images acquired from paediatric patients during surgery.

MATERIALS AND METHODS

The study was performed in compliance with the requirements of the FDA and Boston Children’s Hospital policies. The study was approved by the FDA and institutional IRB (IRB-P00013570) and registered with the National Clinical Trials office (NCT 03189134).

Patients

Patients between the age of 30 days and 21 years undergoing cardiac surgery for secundum atrial septal defect (ASD) repair at Boston Children’s Hospital were the target population. Patients with secundum ASD have normal cardiac physiology with no inherent conduction defects, and the surgical repair has the least risk of postoperative conduction block. This patient selection simplifies the identification of adverse events caused by FCM imaging. Exclusion criteria included prior history of adverse reaction to fluorescein sodium, prior history of renal failure or abnormal renal function, baseline PR interval >220 ms or 98% for age, baseline heart rate >98% for age and any underlying genetic syndrome associated with progressive AV block or SN dysfunction, e.g. Holt–Oram or NKX2.5.

For patient recruitment, a letter with study information was sent to the family ahead of the preoperative visit. Usually, on the day before surgery, subjects and their families were approached in the preoperative clinic by a research nurse. The family was given a copy of the consent form and the research nurse reviewed the study in lay terms. Families were given until the day of surgery to determine if they wished to participate in the study. Written informed consent to participate in the study was then obtained by the research nurse. The consent of both parents was in general obtained. In exceptional cases, consent of a single parent was obtained. The characteristics of the enrolled patients are listed in Table 1.

Table 1:

Patient and operative characteristics

Number	6
Age, n (%)
30 days to 11 years	5 (83)
12–17 years	1 (17)
18–21 years	0
Sex, n (%)
Male	2 (33)
Female	4 (67)
Cross-clamp time (min), mean ± SD	34 ± 5
Dye volume used (ml), mean ± SD	1 ± 0.5

SD: standard deviation.

Technique

We utilized a commercially available Cellvizio 100 series system (Mauna Kea Technologies, Paris, France) with a GastroFlex UHD miniprobe (Fig. 1A–C). The FCM system consisted of 2 main functional parts managed separately in the OR: (i) a movable tower and (ii) detachable miniprobe. The tower houses the confocal laser scanning unit, control-computer, real-time imaging display and control inputs, i.e. keyboard, remote and foot pedal. The laser scanning unit includes a 15-mW 488-nm wavelength laser. The miniprobe transfers laser light to the tissue and the captured fluorescence back to the laser scanning unit. We applied a miniprobe capable of imaging a 240-µm diameter field of view at a depth of 65 µm and with a lateral resolution of 1.8 µm. Imaging was performed at a rate of 9 Hz. Fluorescein sodium dye (Fluorescite^® 10%; Alcon Laboratories Inc., Fort Worth, TX, USA) at 1 mg/ml concentration (1:1000 dilution of Fluorescite) was utilized for topical application for imaging. Fluorescein sodium is an FDA-approved dye commonly utilized for ophthalmic vascular imaging. Fluorescein sodium is a small hydrophilic molecule (molecular weight: 332 g/mol) that readily diffuses within cardiac tissue. Shortly after application, fluorescein sodium distributes in the extracellular space but not in the intracellular regions, revealing the tissue microstructure.

Figure 1:

FCM system and operating room schematic. (A–C) Components of the FCM system. (A) The movable tower consisting of the display on top with a keyboard and remote footswitch for operating the microscope. Beneath the display and keyboard are the laser scanning unit and processing computer. The footswitch is stored in the lowest shelf on the tower. (B) A GastroFlex UHD miniprobe is ∼3-m long with a connector on the proximal end that attaches to the laser scanning unit. (C) The distal end of probe consists of a small objective, which is manoeuvred for imaging. (D) Schematic representation of the cardiac operating room showing the position of the FCM movable tower and the routeing of the miniprobe to the surgeon. FCM: fibre-optic confocal microscopy.

The FCM tower was moved into the OR, turned on, warmed up and calibrated prior to surgery. The miniprobes were stored separately, processed and managed by the surgical instrument processing department and handed to the OR circulation nurse before surgery. Fluorescite was diluted 1:1000 with saline in sterile conditions and loaded into a 5-ml syringe by the hospital pharmacy. Up to 5 ml of dye could be used as needed by the surgeon to produce good image contrast.

After the sternotomy, the subject was placed on cardiopulmonary bypass. After cardioplegic infusion and cardiac arrest, the right atrium was opened to visualize the ASD. The nodal regions were identified based on the established anatomical landmarks. The imaging team then applied the fluorescent dye and imaged the SN region for 60 s, followed by the AVN region for 60 s and then the right ventricular myocardium for 30 s. Afterward, imaging was stopped. The heart surface was washed with saline and evacuated using suction to ensure all fluorescein sodium had been removed and did not enter the circulatory system. The subject was monitored at the time of dye application for any signs of allergic reaction or intolerance. The total FCM imaging time was maintained under 3 min for each surgical procedure. After imaging, the remainder of the operation proceeded in the usual manner. All FCM video clips were saved during intraoperative FCM imaging, and the corresponding anatomical locations were noted.

Subjects received standard care in the postoperative period. The principal investigator and research team monitored clinical laboratories, electrocardiograms and echocardiograms for any evidence of adverse events. All adverse events were recorded for the subjects until discharge from the hospital. The adverse events were classified according to NCI-CTCAE v 4.03 [27]. Adverse events were recorded and followed until each event was resolved or medically stable, or until the subject was discharged from the hospital, whichever came first. The principal investigator and a clinical nurse reviewed the adverse events and determined if they were associated with the use of fluorescein sodium and/or the FCM imaging. These determinations were further reviewed by the study’s independent Data and Safety Monitoring Board (DSMB).

Expert fibre-optic confocal microscopy image review

FCM video clips collected from the patients were de-identified and randomized. FCM images were extracted from the video clips for FCM image review. A first reviewer selected 60 FCM images and classified them as reticulated, striated or of poor quality (indistinguishable microstructure) with 20 images in each group. Eight trained reviewers, blinded to the image classification by reviewer 1, reviewed the selected FCM images. The reviewers classified if images had striated, reticulated or indistinguishable microstructures. The responses of the reviewers were then used to determine the reliability of agreement by computing Fleiss’ kappa (Matlab 2018a; Mathworks, Natick, MA, USA) [28].

Statistical analysis

Mean and standard deviations were calculated where appropriate.

RESULTS

During the study, 7 subjects were consented and only 6 were enrolled. One subject was withdrawn due to the surgeon’s unavailability. The 6 enrolled subjects all completed the study.

Fibre-optic confocal microscopy system and fluorescein sodium safety

All adverse events experienced by the study subjects were determined to be expected for this patient population and unrelated to study treatment. There were no rhythm events, bradycardia or adverse effects on the conduction system. Table 2 lists all adverse events experienced by the study subjects.

Table 2:

Summary of adverse events

Body system	Adverse event type	Severity (grade)	n	Incidence (%)
Circulatory	Anaemia	1	1	16.7
Circulatory	Anaemia	2	2	33.3
Circulatory	Anaemia	3	2	33.3
Circulatory	Hypertension	1	1	16.7
Circulatory	Oedema	1	2	33.3
Digestion	Nausea	1	2	33.3
Digestion	Nausea	2	1	16.7
Digestion	Constipation	1	4	66.7
Digestion	Constipation	2	2	33.3
Digestion	Emesis	2	2	33.3
Nervous	Pain	1	1	16.7
Nervous	Pain	2	5	83.3
Nervous	Insomnia	1	1	16.7
Nervous	Agitation	1	1	16.7
Respiratory	Pneumothorax	1	2	33.3
Respiratory	Pneumothorax	2	1	16.7
Respiratory	Pleural effusion	1	1	16.7
Respiratory	Pleural effusion	2	1	16.7
Respiratory	Atelectasis	1	2	33.3
Respiratory	Atelectasis	2	2	33.3

Fibre-optic confocal microscopy system or integration

The FCM tower and imaging protocol were seamlessly integrated into the OR without affecting workflow. The tower was moved in after the patient had been anesthetized for surgery and positioned opposite the surgeon with the display screen turned to his/her sight (Fig. 1D). There were 2 minor protocol deviations regarding total study time of dye administration and FCM imaging, which were managed by a change in workflow and did not affect the efficacy of imaging. Also, this did not affect the total cross-clamp time for the first 2 patients.

Image acquisition using fibre-optic confocal microscopy system and blinded review

Videos 1 and 2 present FCM video clips captured during imaging of the AVN region. Video 1 shows a reticulated microstructure with the miniprobe placed within the AVN region. Video 2 shows the transition from striated to reticulated microstructure as the miniprobe was manoeuvred from the right atrium (RA) to the AVN region within the triangle of Koch. Microstructures captured during FCM imaging at SN region (Fig. 2B and C) and AVN region (Fig. 2E and F) were reticulated, while those captured in the atrial working myocardium (Fig. 2A and D) region were striated. These findings are in agreement with results from our previous studies [20, 24, 25]. During imaging, <1 ml of fluorescein sodium in sterile saline applied topically was found to be sufficient to produce good imaging contrast, as opposed to the allowed 5 ml.

Figure 2:

Fibre-optic confocal microscopy images from paediatric patient. Striated microstructures captured during imaging working myocardium while approaching the (A) sinus node and (D) atrioventricular node regions. Reticulated tissue microstructures captured during imaging in the (B and C) sinus node region and (E and F) atrioventricular node region (scale bar: 20 μm).

A set of 60 FCM images were evaluated by blinded reviewers. Reviewer responses are tabulated in Supplementary Material, Table S3. The responses yielded a Fleiss’ kappa of 0.822. The results demonstrate a high agreement of classifications among reviewers (Table 3).

Table 3:

Inter-observer agreement on image classification

Reviewer	Inter-observer agreement (%)
	1	2	3	4	5	6	7	8	9	Average	Range
1		90.0	96.7	91.7	93.3	81.7	91.7	90.0	88.3	90.4	81.7–96.7
2	90.0		93.3	88.3	93.3	81.7	88.3	90.0	81.7	88.3	81.7–93.3
3	96.7	93.3		95.0	96.7	85.0	95.0	93.3	85.0	92.5	85.0–96.7
4	91.7	88.3	95.0		91.7	80.0	90.0	91.7	86.7	89.4	80.0–95.0
5	93.3	93.3	96.7	91.7		81.7	91.7	93.3	81.7	90.4	81.7–96.7
6	81.7	81.7	85.0	80.0	81.7		83.3	81.7	73.3	81.0	73.3–85.0
7	91.7	88.3	95.0	90.0	91.7	83.3		88.3	83.3	89.0	83.3–95.0
8	90.0	90.0	93.3	91.7	93.3	81.7	88.3		85.0	89.2	81.7–93.3
9	88.3	81.7	85.0	86.7	81.7	73.3	83.3	85.0		83.1	73.3–88.3

DISCUSSION

This first human clinical study using FCM during cardiac surgery demonstrated that the FCM system can be seamlessly integrated into the OR and surgical workflow during heart surgery. In addition, the use of FCM imaging along with the application of fluorescein sodium did not result in complications or adverse events that were attributed to the introduction of the new imaging approach. Importantly, we found that reviewers can reliably classify acquired FCM images, which we previously showed provide a foundation for localization of the cardiac conduction system in rodent and ovine models [25, 26].

Currently, the AVN is localized during surgery by identifying the triangle of Koch [9]. The triangle is formed by the base at the orifice of the coronary sinus, the tendon of Todaro and the annulus of the septal leaflet of the tricuspid valve [9]. The triangle of Koch localizes not only the central AVN but also other nodal sections [29, 30]. The AVN lies adjacent to the central fibrous body. A transitional zone separates the compact node and atrial overlay spanning from the atrial working myocardium towards the tricuspid valve. As the node progresses along the annulus of the tricuspid valve, it forms the His bundle, which is surrounded by fibrous tissue to insulate it from working myocardium. The His bundle branches into the left and then the right bundle branches. Our FCM system acquired images at a depth of 65 µm and thus reticulated tissues imaged in the AVN region could be from structures superficial to the node. Detailed characterization of the imaged region would require techniques such as histology, which was not feasible in our clinical trial. The SN, on the other hand, is known to be more superficial and includes similar reticulated microstructural features when imaged with FCM. These differences in the depths of these 2 nodes may lead to differences in the accuracy of the precise nodal location. However, our results indicate that both regions yield microstructural features indicative of nodal region location. Therefore, while FCM may not be able to image the node directly, the microstructural differences between nodal regions and working myocardium are distinct such that a user can identify general regions of the AVN and SN. This information can hopefully be utilized to avoid trauma or injury to these regions.

Limiting the total imaging time to 3 min during this clinical trial led to only a small number of images of tissue microstructures. This is explained in part by the difference in surgical access to the heart in humans versus access in ovine heart [25]. This difference required the surgeon to modify the miniprobe manoeuvring strategy during the human trial. It is vital for surgeons to gain sufficient training in probe handling for high-quality image acquisition. We anticipate that, as surgeons become more experienced with the FCM system, efficiency and reliability of image acquisition will increase.

In general, imaging started in near-nodal regions without distinguishable and/or with characteristic myocardial microstructure and stopped when reticulated patterns were clearly visible. Intraoperatively any clinical decision would be made only on distinguishable images. It is important for the surgeon to avoid decisions based on poor-quality images. Hence, we included a separate group of FCM images without distinguishable microstructure in our reviewer analyses and demonstrated that these images can be reliably classified by reviewers.

The use of anatomical landmarks for localizing nodal regions has been fairly well established in hearts with a normal structure or common congenital defects [9, 12, 15]. The rate of postoperative conduction block has not changed significantly in the last 2 decades [4], indicating a saturating point in the effectiveness of this approach. The complication rate varies across the spectrum of defect types with higher rates in the more complex cases. In ASD patients as in our study, conduction tissue regions are reliably identifiable. However, in more complicated congenitally deformed hearts, these conduction tissue regions may be absent or displaced, and thus more difficult to identify. During surgery, the surgeon must estimate the location of the regions using knowledge from published histological studies on the specific defect. Compared to the use of anatomical landmarks, the use of FCM for intraoperative identification of conduction tissue regions could supplement detailed knowledge of cardiac anatomy. We anticipate that the primary application for this technique in cardiac surgery will be to inspect probable nodal regions before making incisions or placing sutures. Another potential benefit would be to reduce residual lesions by inspecting nodal regions using the FCM system. In complex cases such as atrial isomerism/heterotaxy syndrome [17] and congenitally corrected transposition [16], the location of the conduction system has significant variability. The real-time imaging system that we described may help guide intraoperative repair in these lesions and could potentially reduce the incidence of postoperative conduction delays.

In our clinical procedures, we included an imaging assistant in the OR. This person was responsible for managing the FCM tower and for initiating the imaging sequence under verbal command from the surgeon. The imaging system includes a footswitch to initiate the imaging sequence. However, this switch is not optimized for cardiac surgery. To use the FCM system for this and future near-term studies, we could not alter the FDA-approved FCM system by incorporating another type of switch mechanism. We rather decided to include another person to manually start the imaging sequence when requested by the surgeon. We anticipate that future design improvements to the imaging probe include an electrocautery hand wand-like switch, which will allow the surgeon to control the image acquisition and presentation.

Our assessment of feasibility and safety shows promise for FCM imaging as an intraoperative tool for tissue localization in paediatric cardiac surgery. Our upcoming studies are designed to determine the effectiveness of our presented technique in reducing postoperative conduction delays and residual lesions. These studies will also provide insights into the costs of the imaging approach, which we expect to be small versus the large costs of rhythm management due to surgical complications.

CONCLUSIONS

Our studies constitute an important step towards clinical translation of FCM imaging. The FCM system using fluorescein sodium dye was found to be safe for use during congenital heart surgery. The FCM system was integrated into the OR and seamlessly utilized during open-heart surgery. The reliability of reviewer agreement on the classification of FCM images was high. Thus, our pilot study on a group of paediatric patients demonstrates the potential for FCM to be an effective tool in locating conduction tissue regions during congenital heart surgery.

SUPPLEMENTARY MATERIAL

Supplementary material is available at EJCTS online.

ABBREVIATIONS

ASD

Atrial septal defect

AVN

Atrioventricular node

FCM

Fibre-optic confocal microscopy

FDA

Food and Drug Administration

OR

Operating room

RV

Right ventricle

SN

Sinus node

Presented at the 33rd Annual Meeting of the European Association for Cardio-Thoracic Surgery, Lisbon, Portugal, 3–5 October 2019.