Assessment of MMP-2/-9 expression by fluorescence endoscopy for evaluation of anastomotic healing in a murine model of anastomotic leakage

Abstract

Background: Disturbance of intestinal wound closure leads to insufficient anastomotic healing and is associated with considerable morbidity following colorectal resections. Matrix metalloproteinases (MMPs) play a crucial role in regulation of wound closure. Here fluorescence endoscopy was evaluated for assessment of MMP-2/-9 expression during failed intestinal anastomotic healing.

Methods: Distal colonic anastomoses were performed as a model for disturbed healing in 36 Balb/c mice. Healing was evaluated endoscopically, macroscopically, and histologically after 1, 3 and 5 days. For detection of MMP-2/-9 expression fluorescence endoscopy (FE) was used following i.v.-administration of a Cy5.5-labeled MMP-2/-9 specific tracer. FE was complemented by quantification of the fluorescence signal using the MS-FX PRO Optical Imaging System. An overall leakage score was calculated and correlated with the results of FE.

Results: With increasing incidence of anastomotic leakage from POD1 (17%) to POD5 (83%) the uptake of the MMP tracer gradually increased (signal-to-noise ratio (SNR), POD1: 17.91 ± 1.251 vs. POD3: 30.56 ± 3.03 vs. POD5: 44.8 ± 4.473, P<0.0001). Mice with defective anastomotic healing showed significantly higher uptake compared to non-defective (SNR: 37.37± 3.63 vs. 26.16± 3.635, P = 0.0369). White light endoscopy and FE allowed evaluation of anastomotic healing and visualization of mucosal MMPs in vivo. Using FE based detection of MMPs in the anastomosis, an overall positive predictive value of 71.4% and negative predictive value of 66.6% was calculated for detection of anastomotic leakage.

Conclusion: During disturbed anastomotic healing increased expression of MMP-2/-9 was observed in the anastomotic tissue. Fluorescence endoscopy for detection of MMP-2/-9 during the healing process might be a promising tool for early identification of anastomotic leakage.

Fig 1. For analysis of anastomotic healing a distal colonic end-end anastomosis was performed.

The small bowel was moved to the side and a distal colonic segment at the recto-sigmoid junction was dissected preserving the vascular arcades of the large bowel. A standardized end-to-end anastomosis was performed using 8 single stitches of 7–0 absorbable suture material, each with 4 full thickness single stitches of the anterior and posterior wall.

Fig 2. The chosen model serves as a valid tool for analysis of anastomotic leakage.

(A) Representative images of endoscopic evaluation of anastomotic healing, anastomosis marked by dotted line, arrows indicate dehiscence. (B) Representative images of histological sections of the anastomoses on the respective postoperative days. Scale bar 150 μm (C) Endoscopic scoring of visual signs of leakage revealed significantly higher scores on POD5 compared to POD1 (p = 0.0428). (D) Histological inflammation scores were significantly higher with pronounced anastomotic inflammation on the later postoperative days (p = 0.0006). (E) An anastomotic leakage score was calculated taking all parameters into account. As shown in the graph, leakage rates significantly increased on the later postoperative days (p = 0.0047).

Fig 3. MMP expression increases during anastomotic healing and correlates with anastomotic leakage.

(A) The photoprobe AF443-Cy5.5 was injected 24h prior to analysis and was detected with significantly increased intensity on the later postoperative days (SNR, POD5 vs. POD1, p <0.0001). (B) Using immunostaining MMP-9 expression can be localized within the anastomotic tissue as well as the perianastomotic inflammatory exudate with increased expression beginning on POD3. Scale bar 150 μm. (C) Immunostaining was corroborated using gelatin zymography. As shown in the gel, elevated expression and activity of MMP-2 and -9 can be detected beginning on POD3. (D) In comparison within the leakage groups (0 = no leakage; 1 = leakage) the leakage group showed significantly higher uptake of the tracer in ex vivo analysis (SNR: vs. 37.37± 3.63 vs. 26.16 ± 3.635, p = 0.0369).

Fig 4. Specificity experiment to rule out unspecific binding of the tracer at the site of inflammation.

Left panel: Ex vivo detection of the specific (Cy5.5-AF443) and unspecific (Cy3.5-Glycine) tracers using the IVIS spectrum system within the same mice was used for investigation of specificity. Right panel: Comparison on the different time points shows significantly higher signal to noise ratio (SNR) with the specific tracer compared to the non specific fluorophore (p = 0.0019).

Fig 5. Fluorescence endoscopy can be used to detect MMP expression for analysis of anastomotic healing in vivo.

(A) Fluorescence endoscopy (FE) was used for detection of the photoprobe AF443-Cy5.5 in the anastomosis 24h following i.v. administration of the tracer. Areas with signs of anastomotic inflammation showed higher fluorescence signals than the rest of the anastomosis (dotted line marking the anastomotic ring). All anastomoses were analyzed and categorized into two groups as either positive (group 1) or negative (group 0). Detection of the photoprobe in vivo well correlated with increase in uptake of the tracer in ex vivo analysis (panel in the middle). (B) Comparison of the two groups revealed significantly higher uptake of the tracer in the group with positive signal in endoscopic analysis (p = 0.0016).