1.

Introduction

For patients with lung cancer, advanced image-guided radiation therapy (RT) could significantly improve clinical outcomes by controlling the lung tumor with radiation dose escalation or acceleration.¹ Lung cancer RT is, however, hampered by toxicity to nearby healthy organs, such as the esophagus, which is particularly vulnerable to irradiation. Acute radiation-induced esophageal damage (ARIED) or acute esophagitis can be a dose-limiting factor during head and neck cancer and lung cancer RT.^2–7 Acute radiation-induced damages appear within three months after RT, and late radiation-induced damages occur more than three months postirradiation.^8,9 ARIED arises in most patients treated with lung and thoracic RT since the gastrointestinal tract is often close to the primary tumor or tumor-bearing lymph nodes.^5,10–12 ARIED may result in reduced food intake, feeding tube requirement, fistula formation, hospitalization, and surgical intervention. These complications are often dose-limiting, necessitating lowering treatment dose and as such hamper the local tumor control.^13,14 Improved knowledge of the effects of irradiation on the organs at risk is paramount to allow better optimization of the balance between tumor control and damage to the neighboring organs. Currently, there are limited means to determine ARIED, such as white light endoscopy (WLE) and positron emission tomography (PET).¹⁵ The presumed relation between the dose distribution and patient symptoms is so far mainly based on toxicity scored in clinical practice, without visualizing ARIED.^13,16 One study found that the expansion of the esophagus on computed tomography (CT) images has potential as an objective measure of toxicity.¹⁴ If we could detect and monitor ARIED, we would be able to modify RT and, thereby, prevent complications avoiding treatment interruptions, which would enhance the quality of life of patients.

Optical coherence tomography (OCT) is a minimally invasive depth-resolved imaging modality to obtain cross-sectional images with high resolution ($10\mu \mathrm{m}$). OCT acquires backscattered near-infrared light from tissue. The depth of OCT imaging is limited to 2.0 to 3.0 mm because of light–tissue interaction.^17,18 Cylindrical catheters with single rotating optical fibers are capable of scanning the esophageal wall surface over a length of 6 cm.^19–24 A probe-based OCT imaging session can be combined with a standard WLE procedure for minimally invasive diagnosis. More recently, a tethered capsule endomicroscopy technique was integrated in a pill-like imaging device meant for voluntary swallowing, which is capable of cross-sectional OCT imaging of the whole esophagus without sedation and minimal burden to patients.^25–27 In contrast to endoscopy, OCT creates three-dimensional (3-D) depth-resolved architectural microscopic images that may be spatially correlated with histopathology. We hypothesize that OCT may detect radiation-induced damage in the esophagus, e.g., by quantifying changes in light scattering of tissue. Cellular changes, such as apoptosis, necrosis, and inflammation postirradiation, are known to change light scattering properties.^28,29

In this study, we qualitatively and quantitatively investigated the feasibility of OCT to detect ARIED preclinically in mice. We compared toxicity based on histopathology reports with OCT findings.

2.

Materials and Methods

2.4.

Methods

2.4.1.

Colocalization of OCT images with histopathology

We had four dose locations; however, because of the difficulties to define these exact locations in the histopathology specimen, we divided the esophageal specimen into three anatomical regions. We subsequently also divided the sagittal OCT esophageal wall image into these anatomical regions and visually matched the histopathology results to the OCT images. The trachea bifurcation was used as anatomical landmark to align 3-D OCT images.

2.4.2.

Data analysis and statistical analysis

Guided by the histopathology results obtained at three months postirradiation, we visually inspected the 3-D OCT images individually to detect esophageal damages. Esophageal damages were identified by changes in tissue scattering properties and layer thickness. Registration of the OCT and the histopathology is qualitative, based on the location along the length of the esophagus. We used in-house developed software (Worldmatch)³¹ for the visualization and processing of our data to better detect esophageal layers and damages. We evaluated the visibility of such damage quantitatively by calculating the contrast-to-background-noise ratio (CNR).³² Refer to the Appendix for a detailed description of the methods.

We also measured the distance between the muscularis mucosa and lamina propria of edema and healthy esophageal tissues to evaluate the visibility of edema quantitatively. We used repeated measures analysis of variance (ANOVA) to evaluate the differences in distances and CNR values (Appendix).

2.4.3.

Experimental protocol

We randomly divided 30 SPF FVB mice into five experimental groups (one test and four control groups), irradiated them once, and imaged for three months. The OCT-only group was compared with the radiation-only group to differentiate between probe insertion- and radiation-related damages in the esophagus. The single-OCT-imaging group of two mice was used to check if single-OCT probe insertion may cause significant damage.

After applying anesthesia, we held the mice by the scruff of the neck in one hand while gently inserting the OCT probe to a standardized level with the other hand. To avoid damaging the tongue, the tongues of the mice were extended and visible during insertion. We advanced the OCT probe down to the stomach opening, where we sensed a bit of an impasse. Inserting the OCT probe up to standardized length marking reduced the risk of probe-induced damage or incomplete tissue coverage with too deep or too shallow insertions, respectively. The OCT probe was disinfected with ethanol after each insertion.

The mice (test, radiation-only, and single-OCT-imaging groups) underwent cone-beam CT imaging for initial setup assessment and dose planning, followed by a single-dose delivery of 4.0, 10.0, 16.0, and 20.0 Gy on four 5.0-mm spots, spaced 10.0 mm apart (Figs. 1 and 2), with the higher doses delivered distally. Irradiation used two orthogonal circular beams at 45 deg. The imaging dose was about 0.7 and 0.5 cGy on the skin and at a 10.0-mm depth, respectively. The OCT probe, preloaded with CT contrast diluted $1:3$ in water, was inserted prior to cone-beam CT scanning, helping to localize the esophagus in cone-beam CT images for treatment planning and dose delivery (Fig. 2).

Fig. 1

Flowchart demonstrating all the steps for the mice (a) in the test group and (b) in the control, radiation-only, single-OCT-imaging, and OCT-only groups.

Fig. 2

(a–c) These cone-beam CT images illustrate the visibility of the OCT probe filled with diluted CT contrast in coronal, sagittal, and axial planes, respectively. (d) A 3-D reconstruction of the cone-beam CT shows the visibility of the OCT probe and illustrates the dose planning. (e–g) These images show the dose distribution (white arrows) in coronal, sagittal, and axial planes overlaid on the planning cone-beam CT, respectively.

Three months postirradiation, each mouse was euthanized, and we dissected the esophagus and stomach attached together. Having the stomach in the specimen helped us to identify proximal and distal esophagus locations and better decide where to cut the tissue for histopathology analysis.

3.

Results

3.1.

Histopathology Findings

Figure 3 demonstrates esophageal damages at three months postirradiation in histopathology and OCT. Table 1 summarizes the histopathology results of all the mice in proximal, middle, and distal esophagus locations for all experimental groups. We found inflammatory infiltration in the test group (4 out of 15 mice) and the radiation-only group (one out of five mice). No incidence of inflammatory infiltration was found in the OCT-only, single-OCT imaging, and control groups. Edema was seen in the test group (9 out of 15 mice) and the OCT-only group (2 out of 5 mice). No incidence of edema was found in the radiation-only, single-OCT imaging, and control groups.

Fig. 3

Histopathology and OCT results showing ARIED in mice. Specimens were cut sagittally: (a) a longitudinal section of the mouse esophagus indicating proximal, middle, and distal parts, (b) focal inflammatory infiltration, (c) multifocal inflammatory infiltration, (d) dyskeratotic and degenerative cells, (e) edema, and (f) inflammatory infiltration and hypertrophy of the endothelial cell in the submucosa.

Table 1

Histopathology results showing ARIED in mice. Since for the single-OCT-imaging group no ARIED was observed at histopathology, this group is not included in this overview.

Group	Proximal esophagus	Middle esophagus	Distal esophagus
Test
Mouse 1	—	Mildly increased collagen in the submucosa, mainly around small vessels with hypertrophy of the endothelial cells	—
Mouse 2	Inflammatory infiltrations in the submucosa and focally in the mucosa	Inflammatory infiltrations in the submucosa	Multifocal inflammatory infiltrations in the mucosa and submucosa
Mouse 3	—	—	—
Mouse 4	—	Mild edema in the submucosa	Small clusters of the squamous cells show dyskeratotic and degenerative changes and pyknosis at the surface of the mucosa
Mouse 5	Edema in the submucosa	—	Mild edema in the submucosa
Mouse 6	Mild edema with increased amount of collagen in the submucosa	Mild edema in the submucosa with increased amount of collagen around blood vessels	Mild edema in the submucosa with increased amount of collagen around blood vessels
Mouse 7	Degeneration (vacuolization) of the epithelial cells accompanied by the edema and inflammatory infiltrations in the mucosa	Mild edema in the submucosa	Inflammatory infiltrations in the submucosa with increased amount of collagen around blood vessels
Mouse 8	—	Local and mild degeneration (vacuolization) of the epithelial cells with mild inflammatory infiltrations	—
Mouse 9	—	—	Single apoptotic cells in the mucosa with mild increase of collagen in the submucosa
Mouse 10	—	—	Dilation of the blood vessels with mildly increased collagen in the mucosa/submucosa
Mouse 11	—	—	—
Mouse 12	—	—	—
Mouse 13	—	—	—
Mouse 14	—	Edema in the submucosa	Congestion of the blood vessels in the mucosa and submucosa with mild and focal inflammation and edema
Mouse 15	—	—	Locally increased cellularity in the lamina propria of the mucosa
OCT-only
Mouse 1	—	—	—
Mouse 2	—	Dyskeratosis in a single cell	—
Mouse 3	Single apoptotic cell in the mucosa	—	Mild edema in the submucosa
Mouse 4	Mild edema in the submucosa	Mild edema in the submucosa	Mild edema in the submucosa
Mouse 5	Mild edema in the submucosa	Mild edema in the submucosa	Mild edema in the submucosa
Radiation-only
Mouse 1	—	—	—
Mouse 2	—	—	—
Mouse 3	—	—	—
Mouse 4	—	—	Local increased amount of collagen and newly formed blood vessels in the submucosa
Mouse 5	—	Focal inflammatory infiltrations in the mucosa	Mild degenerative changes (vacuolization) of the epithelial cells with mild inflammatory infiltrations in the mucosa and congestion in the submucosa
Control
Mouse 1	—	—	—
Mouse 2	—	—	—
Mouse 3	—	—	Focal dyskeratosis of the epithelial cells and focal inflammatory in the mucosa
Mouse 4	—	—	—
Mouse 5	—	—	—

No mice had to be euthanized at humane endpoint during our experiment. In the test group, 27.2% of the edema was found in the proximal esophagus (at low dose) and 36.4% each in the middle and distal esophagus. In the OCT-only group, 28.6% of the edema incidents occurred in both the proximal and middle esophagus; this was 42.8% in the distal part. In total (i.e., for the test and the OCT-only groups combined), the edema incidence was 27.8%, 33.3%, and 38.9% for the proximal, middle, and distal esophagus, respectively.

Inflammatory infiltration incidents in the radiation-only group were seen equally in the middle and distal esophagus with no incidence in the proximal part. In total (i.e., for the test and the radiation-only groups combined), 50.0% of all inflammatory infiltration incidents were distally located (high dose), with 37.5% and 12.5% of total incidents in the middle (intermediate dose) and proximal (low dose) parts of the esophagus, respectively [Fig. 9(a)]. Because most of the inflammatory infiltration was observed at the highest dose region, we concluded that inflammatory infiltration was a radiation-induced damage, while edema was independent of dose and most likely caused by OCT probe insertions.

3.2.

OCT Findings

We solely looked for edema and inflammatory infiltration damages—as the major histopathology reported damages—in our OCT analysis. Guided by histopathology, we found that areas with inflammatory infiltration had high scattering regions that infiltrated/induced into either the mucosa or the submucosa, i.e., at one certain depth. Our quantitative analyses were solely performed for these regions, excluding high scattering signals elsewhere. Edema mostly had low scattering properties combined with swelling of the submucosal layer. The last OCT imaging time point—three months postirradiation and right before the mice were euthanized—was just before histopathology was performed. We found that 85.7% of the inflammatory infiltration and 81.3% of the edema reported in histopathology were detected in these OCT images.

Figure 4 shows an in vivo endoscopic OCT image of a healthy mouse esophagus that illustrates different layers (epithelium, lamina propria, muscularis mucosa, submucosa, and muscle layers). Figure 5(a) shows a healthy esophagus while Fig. 5(b) shows inflammatory infiltration in the distal esophagus and edema in the middle esophagus. Figure 6 shows en-face view of inflammatory infiltration (high scattering that infiltrates in mucosa or submocusa) and edema (low scattering with swelling) in the proximal and middle esophagus. Figure 7 shows the corresponding findings on inflammatory infiltration and edema in histopathology and OCT images. Figure 8 shows the edema and the inflammatory infiltration of a mouse from the test group at all OCT imaging time-points, to facilitate monitoring damages over time.

Fig. 4

OCT images of an in vivo normal esophageal wall in mice. Esophageal wall layers are indicated as: ep = epithelium, lp = lamina propria, mm = muscularis mucosa, sm = submucosa, and ml = muscle layers. (a) Axial plane (the image is unfolded to appear straight rather than circular), (b) axial plane (the images are averaged over three slices perpendicular to the view direction, to improve signal-to-noise ratio and therewith the visualization), (c) sagittal plane (note that sagittal view has inferior resolution due to the image acquisition procedure), and (d) axial view averaged over three slices perpendicular to the view direction.

Fig. 5

(a) In vivo OCT image of a control mouse that shows no inflammatory infiltration and edema and (b) in vivo image of a mouse from the test group illustrating inflammatory infiltration in the distal esophagus and edema in the middle esophagus. The images are averaged over 21 slices perpendicular to the view direction.

Fig. 6

(a) En-face illustration of an in vivo OCT image of a normal esophagus represents trachea and the bifurcation ($\sim 0.4\text{-}\mathrm{mm}$ depth), (b) same at $\sim 0.2\text{-}\mathrm{mm}$ depth, and (c) illustrates edema in the middle esophagus and inflammatory infiltration in the distal esophagus (high-dose region) at about submucosal depth ($\sim 0.2\text{-}\mathrm{mm}$ depth). The images are averaged over 21 slices perpendicular to the view direction, to improve signal-to-noise ratio and therewith the visualization.

Fig. 7

(a, b) Corresponding inflammatory infiltration damages in the middle esophagus in histopathology and OCT. (c, d) Corresponding edema damages in the same region of the same mouse.

Fig. 8

(a) In vivo OCT images of edema in mouse 6 (test group) at 6 time-points postirradiation. The images are averaged over 21 slices perpendicular to the view direction. Yellow arrows point to the suspected edema in the submucosa. Red arrows indicate the tracheal bifurcation level as reference of the middle esophagus. (b) In vivo OCT images of inflammatory infiltration in mouse 8 (test group) at 6 time-points postirradiation. Yellow arrows point to the suspected inflammatory infiltration incidents. Red arrows indicate the tracheal bifurcation level as the reference of the middle esophagus.

We quantitatively measured the differences between healthy tissue and edema in terms of the distance between the muscularis mucosa and lamina propria on OCT [Fig. 9(b)]. There was a statistically significant difference between the two groups ($p=0.00025$). These distances for edema and healthy esophageal wall on $\text{average}\pm \text{standard deviation}$ were $0.9\pm 0.2\mathrm{mm}$ (range: 0.6 to 1.2 mm, median 1.0 mm) and $1.6\pm 0.5\mathrm{mm}$ (range: 1.1 to 2.7 mm, median 1.6 mm), respectively. On average, the distance between muscularis mucosa and lamina propria was 1.7-fold higher in edema than in the healthy esophageal wall.

Fig. 9

(a) Number of inflammatory infiltration occurrences in histopathology and OCT as a function of dose (proximal, middle, and distal esophagus) after three months, as well as for the OCT control group. (b) These boxplots illustrate the differences in distance between the muscularis mucosa and lamina propria esophageal layers for healthy esophageal tissue (left) and edema (right). Boxes indicate interquartile ranges, horizontal lines illustrate the median values, the circles are outliners, and error bars represent the range. (c) Boxplot of CNR values for healthy esophageal wall, edema, and inflammatory infiltration after three months. (d) The boxplots show CNR values of inflammatory infiltration at several time-points postirradiation.

Our quantitative analysis based on CNR differences between healthy esophageal wall, edema, and inflammatory infiltration at three months postirradiation is summarized in Fig. 9(c). There was a statistically significant difference among the three groups ($p=0.00035$). On average, the CNR was 1.6-fold higher in inflammatory infiltration than in the healthy esophageal wall. The CNR of edema was 1.6-fold lower than the healthy esophageal wall. The CNRs for healthy esophageal wall, edema, and inflammatory infiltration on $\text{average}\pm \text{standard deviation}$ were $17.1\pm 6.2$ (range: 8.7 to 27.6, median 15.1), $10.9\pm 5.0$ (range: 4.0 to 19.5, median 10.4), and $28.0\pm 9.0$ (range: 15.5 to 36.1, median 31.4), respectively.

Moreover, we calculated the CNR values of all inflammatory infiltration incidents observed at a subset of time-points [Fig. 9(d)]. The total $\text{average}\pm \text{standard deviation}$ of the calculated CNRs of the inflammatory infiltration damage was $53.6\pm 41.5$ (range: 15.5 to 193.8, median 38.8), and this average CNR was 1.9-fold higher than the CNR of inflammatory infiltration after three months and 3.1-fold higher than the CNR of the healthy esophageal wall.

4.

Discussion

In this study, we investigated ARIED as a function of dose using four increasing dose levels from the proximal to distal end of the esophagus. After irradiation, we imaged them for three months. Compared with the histopathology results acquired at three months after dose delivery, OCT was capable of detecting inflammatory infiltration and edema in mice as a change in light scattering and esophageal wall layer thickness compared to unirradiated mice. Our results suggest that inflammatory infiltration is a radiation-induced damage, while multiple OCT probe insertions caused edema. Edema caused expansion between the muscularis mucosa and lamina propria esophageal walls. Both changes were small but statistically significant on quantitative analysis. Single-OCT probe insertion was less damaging; hence, we saw no edema in the single-OCT-imaging group. The OCT probe insertion-induced edema should be much lower or absent in humans due to availability of dedicated esophageal OCT catheters that can be inserted in a deflated state³² and because of the larger size of the organ.

Previous studies investigated the use of OCT in esophageal and radiation-induced damages,^20,33–46 the visibility of esophageal layers in OCT images, and the differentiation between healthy and cancerous esophagus.^39–41 Esophageal OCT has been used to detect Barrett’s esophagus in various studies.^20,40,42,43 Some studies used functional OCT microvascular imaging for radiobiological monitoring of different radiation dose levels in pancreatic human tumor xenografts⁴⁷ and in head and neck cancer using multifunctional OCT in a clinical pilot study in patients.⁴⁸ Moreover, there are pilot patient studies reported on using microvascular OCT imaging with developed oral imaging probes to monitor late oral radiation-induced damages in vivo.^36,49 However, it is unknown whether we can directly compare these findings with dose response of the esophagus in mice.

Other side effects, such as ulceration, perforation, and fistula formation exist besides the ARIED reported here.^11,50 Edema and fistula formation are late RT complications that increase significantly with dose.⁵¹ Also concurrent chemotherapy increases the risk of radiation-induced damages.^8,50 In this study, we solely focused on the acute dose-related toxicity and we did not use chemotherapy.

Although inflammatory infiltration was the main histopathologically reported incident in OCT images and appeared as a high scattering region, not all high scattering regions are due to inflammatory infiltration. We defined inflammatory infiltration as high scattering regions that infiltrated into the mucosa or submucosa. We found high scattering regions in almost all OCT images, including in the control group, which are not infiltrating into the upper esophageal layers (Fig. 10). We suspect that these regions may represent muscle tissue. There were also regions in OCT that looked like edema or inflammatory infiltration, which were not reported on histopathology analysis, suggesting that we may have undersampled the specimens. Because of the lack of histopathology validation, these regions were not analyzed in OCT images.

Fig. 10

In vivo OCT images of a mouse esophagus showing examples of high scattering areas that are present in almost all cases.

Histopathology results (Table 1) indicated that we only induced limited acute damage in the esophagi of the mice. We hypothesize that the small spot size and chosen dose levels may play a role. Studies in rats have shown that no histopathologic damage is associated with delivered dose up to 18.0 and $<20.0\mathrm{Gy}$ to the cervical spinal cord.^52,53 However, in mice, esophageal radiation-induced damage at one week postirradiation was reported when 30.0 Gy was delivered.⁵⁴ We delivered the maximum single dose of 20.0 Gy in a 0.5-mm spot to limit normal tissue toxicity outside the esophagus, prevent morbidity, and preserve the mice’s physical condition for repeated imaging for three months. In view of uncertainties in the image-guided small animal irradiation system and/or biological effects limiting damage for small regions, our selected spot size may have been too small, and the small fields might also have caused us to miss damaged areas as a result of undersampling of the histopathology. Although for this study it was not feasible to address the undersampling by embedding the entire esophagus for all the mice because of the timely and expensive process, we could revisit this in the future. The assumed human esophageal tolerance for single fraction irradiation is 27 Gy, using an $\alpha /\beta$ ratio of $3{\mathrm{Gy}}^{-1}$.⁵⁵ From our results, using a dose that is below that reported tolerance, we cannot conclude whether the radiobiological sensitivity of mice is different from humans. However, the esophageal wall layering of humans and mice is similar; therefore, our model is a reasonable choice for preclinical investigation. The esophagus is a serial organ⁵⁶ where damage to a single functional subunit may result in damaging the entire organ.⁵⁷ Therefore, the esophagus may show different results, in which case we would have a single irradiation spot compared to our findings for consecutive irradiation spots along the organ. However, modifying treatment parameters, such as the dose distribution, may prevent esophageal strictures.⁴ Our experimental protocol was designed such that we investigated the ARIED with four dose levels per mice, using each mouse as its own control. A limitation of this setup is that we could not study whether there is a serial effect—where organ response is dependent on the length of the irradiated tissue.

Figure 6(a) shows the trachea in an en-face OCT image. The trachea often appeared as a serpentine organ because of nonuniform rotational distortion (NURD).^22,58–61 NURD depends on various parameters, such as catheter specifications and material, subject movement, specific path in the body during the pullback, and the pullback position.⁵⁸ The imaging artifacts caused by NURD, breathing, and probe movements in our study mainly affected the visual quality of the OCT images, resulting in difficulties to interpret some of the data.

Future work includes protocol modification to make sure that we induce more substantial ARIED and perform another small animal study. In future studies, one will be able to reduce the probability of edema formation in the mice by performing less OCT imaging. This can be followed by investigating the feasibility of OCT to detect and monitor ARIED in humans and in vivo in patients with lung cancer undergoing RT using a dedicated esophageal OCT system. We expect that our in vivo findings can be translated to humans because of the availability of dedicated OCT probes and esophageal OCT imaging systems, although damage may occur at other dose levels.^32,40,62

In conclusion, we studied the feasibility of OCT to detect and monitor ARIED in a small animal model (mice). We irradiated the mice using single-dose delivery of 4.0, 10.0, 16.0, and 20.0 Gy on small spots. Inflammation was mostly seen at the highest dose region. Our results indicated the potential role of OCT to assess and monitor the ARIED in mice by detecting changes in light scattering in tissue and esophageal wall layer thicknesses, which may translate to humans.