View Article PDF

Network meta-analysis of antibiotics and bowel preparation in elective colorectal surgery

K Clifford, JC Woodfield, B Schmidt, G Turner, MA Amer, J McCall.
Department of Surgical Sciences, University of Otago, Dunedin.

There are discrepancies in guidelines on bowel preparation for colorectal surgery. While intravenous (IV) antibiotics are commonly administered, the use of mechanical bowel preparation (MBP), enema (E) and/or oral antibiotics (OA) is controversial. This controversy stems in part from the historical use of inadequate IV antibiotics. Our aim was to summarise all data from randomised controlled trials (RCT) by using network meta-analysis (NMA) to determine the ranking of different bowel preparation treatment strategies. This NMA is the first comprehensive review of this topic that accounts for the impact of antibiotic type and compares the efficacy of all bowel preparation options in reducing postoperative infections.

NMA was performed according to PRISMA guidelines. RCT of adult patients undergoing elective colorectal surgery cover were included. The search included Medline, Embase, Cochrane and SCOPUS databases. Primary outcomes were wound infection (SSI) and anastomotic leak (AL). The NMA was performed in Stata v15.1, using three models: all identified studies, studies with appropriate (aerobic and anaerobic) cover when combining IV and OA, and those with appropriate IV antibiotic cover. Sub-analyses were performed using studies examining rectal, left-sided, and right-sided regions of the colon.

We identified 75 RCTs including 16,891 patients. Treatments compared MBP+IV (5,642 patients), IV (2803 patients), IV+E (397 patients), IV+OA± E (649 patients), MBP+IV+OA (4821 patients), MBP+OA (2093 patients) and OA (486 patients). The likelihood of SSI was significantly lower for IV+OA±E (rank 1) and MBP+IV+OA (rank 2) when compared to other treatments (IV, MBP+IV, MPB+OA and OA), both P < 0.001. The addition of OA to IV antibiotics reduced SSI by approximately 50%. There were minimal differences in AL and in secondary endpoints.

This NMA supports the addition of OA to IV antibiotics for patients undergoing elective colorectal surgery. Additional research should assess the effectiveness of IV+OA±E and MBP+IV+OA.

Supported by a University of Otago Research Grant.

A scalable, efficient machine learning pipeline to quantify myocardial collagen in whole slide histological images

L Ariyasinghe, M Moharram, S Coffey.
Department of Medicine, Otago Medical School, Dunedin Campus, University of Otago, Dunedin.

The collagen mesh found in the extracellular matrix (ECM) of the myocardium plays a critical role in preserving tissue architecture and mechanical properties. Dynamic remodelling of the ECM occurs in different myocardial pathologies; this can manifest as a change in chamber geometry and/or mechanical properties which ultimately impacts overall cardiac performance. Therefore, precise quantification of collagen in the myocardium is of particular importance in histological characterisation of myocardial diseases.

Using Masson's trichrome (MTC) to detect collagen is a widely accepted standard in histopathology. However, manual examination of collagen in stained slides can be time-consuming and is non-quantitative, while off-the-shelf computational approaches lack scalability and have rarely been formally validated. Accordingly, we developed a highly scalable and reproducible Machine Learning (ML) pipeline to quantify the amount of collagen in MTC-stained Whole Slide Images (WSIs).

The ML pipeline consists of three distinct phases. 1) Preprocessing—splits the WSI into smaller images (tiles) and removes any tiles with no tissue present. 2) ML inference—applies the ML algorithms, namely, K-Nearest Neighbors (K-NN), Support Vector Machines (SVM) and Random Forest (RF). The ML algorithms were trained on 4567 pixels from manually annotated tiles of right atrial appendage samples, obtained from patients undergoing cardiac surgery. 3) Segmentation—creates a fully segmented tile by classifying each pixel into one of the three classes, specifically, collagen, non-collagen and blank. The results from each tile are then aggregated to provide a value for the WSI as a whole.

We estimated the classification accuracy using a test set of 87200 pixels, obtained from samples from 10 patients. Consequently, reported accuracies are as follows: K-NN—98.3%, SVM—99.6% and RF—98%. In addition to this remarkable accuracy, our ML pipeline provides workflow efficiency and ability to scale up to handle large number of WSIs.

Supported by funding from the Department of Medicine and the New Zealand Heart Foundation.

Structural basis for active-site probes targeting Staphylococcus aureus serine hydrolase virulence factors

M Fellner[[1]], CS Lentz[[2,3]], SA Jamieson[[1]], JL Brewster[[1]], L Chen[[2,4]], M Bogyo[[2]], PD Mace[[1]].
1Biochemistry Department, School of Biomedical Sciences, University of Otago, 2Department of Pathology, Stanford University, USA, 3Department of Medical Biology, The Arctic University of Norway, Norway, 4Shanghai Institute of Materia Medica, Chinese Academy of Sciences, China

Around a third of healthy humans are carriers of Staphylococcus aureus, they have the bacteria on their skin without any active infection or disease. Despite being harmless in most individuals, S. aureus can cause pathogenic infections. It often exists in biofilms in human tissue, resulting in a biomolecular matrix that is largely impermeable to the immune system and many traditional antibiotics. The increased occurrence of community-acquired antibiotic-resistant S. aureus strains, often linked to biofilm formation, is a major health threat, requiring urgent development of new diagnostic and therapy options.

Serine hydrolases are a large family of enzymes that play key roles in bacterial homeostasis and survival at the host-pathogen interface during infection. They play a role in biofilms, contributing to the difficulty of achieving effective treatment. This makes serine hydrolases promising new anti-virulence and anti-infectivity targets.

Activity-based profiling identified a family of serine hydrolases, designated fluorophosphonate-binding hydrolases (Fphs), which contribute to virulence of S. aureus in the biofilm niche. Here we report a structure-function characterization of one of these serine hydrolases, FphF, expressed during biofilm forming conditions. We determined that FphF is a promiscuous enzyme, able to cleave hydrophobic lipid substrates with a range of acyl chain lengths. Using this newly acquired structural data and similarities among the Fph family, we show that other Fph proteins, including FphB which was linked directly to virulence, may have a more well-defined substrate specificity. Our structural and biochemical studies confirm that FphF is distinct from previously characterized enzymes, making it an important reference enzyme in the serine hydrolase superfamily. Overall, our results provide the first insight into the specificity and the mechanism of substrate and chemical probe binding to the Fph protein family. This information will aid in future efforts to targeting serine hydrolase virulence factors from S. aureus and other related bacteria.

This work was supported by a University of Otago Health Sciences Postdoctoral Fellowship.

Summary

Abstract

Aim

Method

Results

Conclusion

Author Information

Acknowledgements

Correspondence

Correspondence Email

Competing Interests

For the PDF of this article,
contact nzmj@nzma.org.nz

View Article PDF

Network meta-analysis of antibiotics and bowel preparation in elective colorectal surgery

K Clifford, JC Woodfield, B Schmidt, G Turner, MA Amer, J McCall.
Department of Surgical Sciences, University of Otago, Dunedin.

There are discrepancies in guidelines on bowel preparation for colorectal surgery. While intravenous (IV) antibiotics are commonly administered, the use of mechanical bowel preparation (MBP), enema (E) and/or oral antibiotics (OA) is controversial. This controversy stems in part from the historical use of inadequate IV antibiotics. Our aim was to summarise all data from randomised controlled trials (RCT) by using network meta-analysis (NMA) to determine the ranking of different bowel preparation treatment strategies. This NMA is the first comprehensive review of this topic that accounts for the impact of antibiotic type and compares the efficacy of all bowel preparation options in reducing postoperative infections.

NMA was performed according to PRISMA guidelines. RCT of adult patients undergoing elective colorectal surgery cover were included. The search included Medline, Embase, Cochrane and SCOPUS databases. Primary outcomes were wound infection (SSI) and anastomotic leak (AL). The NMA was performed in Stata v15.1, using three models: all identified studies, studies with appropriate (aerobic and anaerobic) cover when combining IV and OA, and those with appropriate IV antibiotic cover. Sub-analyses were performed using studies examining rectal, left-sided, and right-sided regions of the colon.

We identified 75 RCTs including 16,891 patients. Treatments compared MBP+IV (5,642 patients), IV (2803 patients), IV+E (397 patients), IV+OA± E (649 patients), MBP+IV+OA (4821 patients), MBP+OA (2093 patients) and OA (486 patients). The likelihood of SSI was significantly lower for IV+OA±E (rank 1) and MBP+IV+OA (rank 2) when compared to other treatments (IV, MBP+IV, MPB+OA and OA), both P < 0.001. The addition of OA to IV antibiotics reduced SSI by approximately 50%. There were minimal differences in AL and in secondary endpoints.

This NMA supports the addition of OA to IV antibiotics for patients undergoing elective colorectal surgery. Additional research should assess the effectiveness of IV+OA±E and MBP+IV+OA.

Supported by a University of Otago Research Grant.

A scalable, efficient machine learning pipeline to quantify myocardial collagen in whole slide histological images

L Ariyasinghe, M Moharram, S Coffey.
Department of Medicine, Otago Medical School, Dunedin Campus, University of Otago, Dunedin.

The collagen mesh found in the extracellular matrix (ECM) of the myocardium plays a critical role in preserving tissue architecture and mechanical properties. Dynamic remodelling of the ECM occurs in different myocardial pathologies; this can manifest as a change in chamber geometry and/or mechanical properties which ultimately impacts overall cardiac performance. Therefore, precise quantification of collagen in the myocardium is of particular importance in histological characterisation of myocardial diseases.

Using Masson's trichrome (MTC) to detect collagen is a widely accepted standard in histopathology. However, manual examination of collagen in stained slides can be time-consuming and is non-quantitative, while off-the-shelf computational approaches lack scalability and have rarely been formally validated. Accordingly, we developed a highly scalable and reproducible Machine Learning (ML) pipeline to quantify the amount of collagen in MTC-stained Whole Slide Images (WSIs).

The ML pipeline consists of three distinct phases. 1) Preprocessing—splits the WSI into smaller images (tiles) and removes any tiles with no tissue present. 2) ML inference—applies the ML algorithms, namely, K-Nearest Neighbors (K-NN), Support Vector Machines (SVM) and Random Forest (RF). The ML algorithms were trained on 4567 pixels from manually annotated tiles of right atrial appendage samples, obtained from patients undergoing cardiac surgery. 3) Segmentation—creates a fully segmented tile by classifying each pixel into one of the three classes, specifically, collagen, non-collagen and blank. The results from each tile are then aggregated to provide a value for the WSI as a whole.

We estimated the classification accuracy using a test set of 87200 pixels, obtained from samples from 10 patients. Consequently, reported accuracies are as follows: K-NN—98.3%, SVM—99.6% and RF—98%. In addition to this remarkable accuracy, our ML pipeline provides workflow efficiency and ability to scale up to handle large number of WSIs.

Supported by funding from the Department of Medicine and the New Zealand Heart Foundation.

Structural basis for active-site probes targeting Staphylococcus aureus serine hydrolase virulence factors

M Fellner[[1]], CS Lentz[[2,3]], SA Jamieson[[1]], JL Brewster[[1]], L Chen[[2,4]], M Bogyo[[2]], PD Mace[[1]].
1Biochemistry Department, School of Biomedical Sciences, University of Otago, 2Department of Pathology, Stanford University, USA, 3Department of Medical Biology, The Arctic University of Norway, Norway, 4Shanghai Institute of Materia Medica, Chinese Academy of Sciences, China

Around a third of healthy humans are carriers of Staphylococcus aureus, they have the bacteria on their skin without any active infection or disease. Despite being harmless in most individuals, S. aureus can cause pathogenic infections. It often exists in biofilms in human tissue, resulting in a biomolecular matrix that is largely impermeable to the immune system and many traditional antibiotics. The increased occurrence of community-acquired antibiotic-resistant S. aureus strains, often linked to biofilm formation, is a major health threat, requiring urgent development of new diagnostic and therapy options.

Serine hydrolases are a large family of enzymes that play key roles in bacterial homeostasis and survival at the host-pathogen interface during infection. They play a role in biofilms, contributing to the difficulty of achieving effective treatment. This makes serine hydrolases promising new anti-virulence and anti-infectivity targets.

Activity-based profiling identified a family of serine hydrolases, designated fluorophosphonate-binding hydrolases (Fphs), which contribute to virulence of S. aureus in the biofilm niche. Here we report a structure-function characterization of one of these serine hydrolases, FphF, expressed during biofilm forming conditions. We determined that FphF is a promiscuous enzyme, able to cleave hydrophobic lipid substrates with a range of acyl chain lengths. Using this newly acquired structural data and similarities among the Fph family, we show that other Fph proteins, including FphB which was linked directly to virulence, may have a more well-defined substrate specificity. Our structural and biochemical studies confirm that FphF is distinct from previously characterized enzymes, making it an important reference enzyme in the serine hydrolase superfamily. Overall, our results provide the first insight into the specificity and the mechanism of substrate and chemical probe binding to the Fph protein family. This information will aid in future efforts to targeting serine hydrolase virulence factors from S. aureus and other related bacteria.

This work was supported by a University of Otago Health Sciences Postdoctoral Fellowship.

Summary

Abstract

Aim

Method

Results

Conclusion

Author Information

Acknowledgements

Correspondence

Correspondence Email

Competing Interests

For the PDF of this article,
contact nzmj@nzma.org.nz

View Article PDF

Network meta-analysis of antibiotics and bowel preparation in elective colorectal surgery

K Clifford, JC Woodfield, B Schmidt, G Turner, MA Amer, J McCall.
Department of Surgical Sciences, University of Otago, Dunedin.

There are discrepancies in guidelines on bowel preparation for colorectal surgery. While intravenous (IV) antibiotics are commonly administered, the use of mechanical bowel preparation (MBP), enema (E) and/or oral antibiotics (OA) is controversial. This controversy stems in part from the historical use of inadequate IV antibiotics. Our aim was to summarise all data from randomised controlled trials (RCT) by using network meta-analysis (NMA) to determine the ranking of different bowel preparation treatment strategies. This NMA is the first comprehensive review of this topic that accounts for the impact of antibiotic type and compares the efficacy of all bowel preparation options in reducing postoperative infections.

NMA was performed according to PRISMA guidelines. RCT of adult patients undergoing elective colorectal surgery cover were included. The search included Medline, Embase, Cochrane and SCOPUS databases. Primary outcomes were wound infection (SSI) and anastomotic leak (AL). The NMA was performed in Stata v15.1, using three models: all identified studies, studies with appropriate (aerobic and anaerobic) cover when combining IV and OA, and those with appropriate IV antibiotic cover. Sub-analyses were performed using studies examining rectal, left-sided, and right-sided regions of the colon.

We identified 75 RCTs including 16,891 patients. Treatments compared MBP+IV (5,642 patients), IV (2803 patients), IV+E (397 patients), IV+OA± E (649 patients), MBP+IV+OA (4821 patients), MBP+OA (2093 patients) and OA (486 patients). The likelihood of SSI was significantly lower for IV+OA±E (rank 1) and MBP+IV+OA (rank 2) when compared to other treatments (IV, MBP+IV, MPB+OA and OA), both P < 0.001. The addition of OA to IV antibiotics reduced SSI by approximately 50%. There were minimal differences in AL and in secondary endpoints.

This NMA supports the addition of OA to IV antibiotics for patients undergoing elective colorectal surgery. Additional research should assess the effectiveness of IV+OA±E and MBP+IV+OA.

Supported by a University of Otago Research Grant.

A scalable, efficient machine learning pipeline to quantify myocardial collagen in whole slide histological images

L Ariyasinghe, M Moharram, S Coffey.
Department of Medicine, Otago Medical School, Dunedin Campus, University of Otago, Dunedin.

The collagen mesh found in the extracellular matrix (ECM) of the myocardium plays a critical role in preserving tissue architecture and mechanical properties. Dynamic remodelling of the ECM occurs in different myocardial pathologies; this can manifest as a change in chamber geometry and/or mechanical properties which ultimately impacts overall cardiac performance. Therefore, precise quantification of collagen in the myocardium is of particular importance in histological characterisation of myocardial diseases.

Using Masson's trichrome (MTC) to detect collagen is a widely accepted standard in histopathology. However, manual examination of collagen in stained slides can be time-consuming and is non-quantitative, while off-the-shelf computational approaches lack scalability and have rarely been formally validated. Accordingly, we developed a highly scalable and reproducible Machine Learning (ML) pipeline to quantify the amount of collagen in MTC-stained Whole Slide Images (WSIs).

The ML pipeline consists of three distinct phases. 1) Preprocessing—splits the WSI into smaller images (tiles) and removes any tiles with no tissue present. 2) ML inference—applies the ML algorithms, namely, K-Nearest Neighbors (K-NN), Support Vector Machines (SVM) and Random Forest (RF). The ML algorithms were trained on 4567 pixels from manually annotated tiles of right atrial appendage samples, obtained from patients undergoing cardiac surgery. 3) Segmentation—creates a fully segmented tile by classifying each pixel into one of the three classes, specifically, collagen, non-collagen and blank. The results from each tile are then aggregated to provide a value for the WSI as a whole.

We estimated the classification accuracy using a test set of 87200 pixels, obtained from samples from 10 patients. Consequently, reported accuracies are as follows: K-NN—98.3%, SVM—99.6% and RF—98%. In addition to this remarkable accuracy, our ML pipeline provides workflow efficiency and ability to scale up to handle large number of WSIs.

Supported by funding from the Department of Medicine and the New Zealand Heart Foundation.

Structural basis for active-site probes targeting Staphylococcus aureus serine hydrolase virulence factors

M Fellner[[1]], CS Lentz[[2,3]], SA Jamieson[[1]], JL Brewster[[1]], L Chen[[2,4]], M Bogyo[[2]], PD Mace[[1]].
1Biochemistry Department, School of Biomedical Sciences, University of Otago, 2Department of Pathology, Stanford University, USA, 3Department of Medical Biology, The Arctic University of Norway, Norway, 4Shanghai Institute of Materia Medica, Chinese Academy of Sciences, China

Around a third of healthy humans are carriers of Staphylococcus aureus, they have the bacteria on their skin without any active infection or disease. Despite being harmless in most individuals, S. aureus can cause pathogenic infections. It often exists in biofilms in human tissue, resulting in a biomolecular matrix that is largely impermeable to the immune system and many traditional antibiotics. The increased occurrence of community-acquired antibiotic-resistant S. aureus strains, often linked to biofilm formation, is a major health threat, requiring urgent development of new diagnostic and therapy options.

Serine hydrolases are a large family of enzymes that play key roles in bacterial homeostasis and survival at the host-pathogen interface during infection. They play a role in biofilms, contributing to the difficulty of achieving effective treatment. This makes serine hydrolases promising new anti-virulence and anti-infectivity targets.

Activity-based profiling identified a family of serine hydrolases, designated fluorophosphonate-binding hydrolases (Fphs), which contribute to virulence of S. aureus in the biofilm niche. Here we report a structure-function characterization of one of these serine hydrolases, FphF, expressed during biofilm forming conditions. We determined that FphF is a promiscuous enzyme, able to cleave hydrophobic lipid substrates with a range of acyl chain lengths. Using this newly acquired structural data and similarities among the Fph family, we show that other Fph proteins, including FphB which was linked directly to virulence, may have a more well-defined substrate specificity. Our structural and biochemical studies confirm that FphF is distinct from previously characterized enzymes, making it an important reference enzyme in the serine hydrolase superfamily. Overall, our results provide the first insight into the specificity and the mechanism of substrate and chemical probe binding to the Fph protein family. This information will aid in future efforts to targeting serine hydrolase virulence factors from S. aureus and other related bacteria.

This work was supported by a University of Otago Health Sciences Postdoctoral Fellowship.

Summary

Abstract

Aim

Method

Results

Conclusion

Author Information

Acknowledgements

Correspondence

Correspondence Email

Competing Interests

Contact diana@nzma.org.nz
for the PDF of this article

Subscriber Content

The full contents of this pages only available to subscribers.

LOGINSUBSCRIBE