Article Text

Download PDFPDF
A randomised trial of sheathed versus standard forceps for obtaining uncontaminated biopsy specimens of microbiota from the terminal ileum
  1. Maneesh Dave1,
  2. Laura A Johnson2,
  3. Seth T Walk3,
  4. Vincent B Young3,
  5. Ryan W Stidham2,
  6. Meghana N Chaudhary2,
  7. Jessica FunNell2,
  8. Peter D R Higgins2
  1. 1Division of Gastroenterology and Hepatology, Department of Internal Medicine, Mayo Clinic, Rochester, Minnesota, USA
  2. 2Division of Gastroenterology and Hepatology, Department of Internal Medicine, University of Michigan School of Medicine, Ann Arbor, Michigan, USA
  3. 3Division of Infectious Diseases, Department of Internal Medicine and Department of Microbiology and Immunology, University of Michigan School of Medicine, Ann Arbor, Michigan, USA
  1. Correspondence to Assistant Professor Peter Higgins, Division of Gastroenterology and Hepatology, Department of Internal Medicine, University of Michigan Medical Center, SPC 5682, Room 6510D, Medical Science Research Building One, 1150 West Medical Center Drive, Ann Arbor, MI 48109-0682, USA; phiggins{at}med.umich.edu

Abstract

Background The study of intestinal microbiota has been revolutionised by the use of molecular methods, including terminal restriction fragment length polymorphism (T-RFLP) analysis. Microbiota studies of Crohn's disease patients have examined samples from stool or from the neoterminal ileum with a standard biopsy forceps, which could be contaminated by colonic bacteria when the forceps passes through the colonoscope channel.

Objective To determine whether sheathed biopsy forceps are able to obtain terminal ileal microbiota samples with less colonic bacterial contamination compared with unsheathed (standard) biopsy forceps.

Design Prospective randomised single-centre study.

Patients and methods Four (paired) biopsy specimens were obtained from adjacent locations in the terminal ileum using the sheathed and standard forceps of 27 consecutive subjects undergoing colonoscopy and the microbiota were characterised using T-RFLP. The Bray–Curtis similarity index between samples (sheathed vs unsheathed forceps) was calculated within patients and significant differences were tested for across all patients.

Results There was not a significant difference in the microbial diversity of samples obtained using sheathed versus unsheathed forceps. The difference in microbial diversity between patients was much greater than the variability within patients by proximal versus distal site or by forceps type.

Limitations T-RFLP is based on PCR amplification, so it is not always sensitive to rare bacterial species.

Conclusion Standard unsheathed forceps appear to be sufficient for microbiota sample collection from the terminal ileum.

  • Crohn's disease
  • endoscopic procedures
  • inflammatory bowel disease
  • intestinal microbiology
  • microbial diversity
  • microbiota
  • terminal ileum
  • terminal restriction fragment length polymorphism

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Significance of this study

What is already known about this subject?

  • The microbiota of the ileum appears to influence Crohn's disease, as diversion or antibiotics can delay recurrence after surgery.

  • The microbiota of intestinal mucosa is distinct in different segments of the intestine, and distinct from the microbiota of the stool.

  • Treatments to delay the recurrence of Crohn's may first affect the microbiota of the ileum, but sampling the microbiota via a colonoscope may be contaminated by colonic bacteria.

What are the new findings?

  • A sheathed standard forceps can be used to sample the terminal ileal microbiota for molecular analysis.

  • There was not a significant difference in the microbial diversity of samples obtained using sheathed versus unsheathed forceps.

  • The difference in microbial diversity between patients was much greater than the variability within patients by proximal versus distal site or by forceps type.

  • Standard unsheathed forceps appear to be sufficient for microbiota sample collection from the terminal ileum.

How might it impact on clinical practice in the foreseeable future?

  • Evaluation of the microbiota of the terminal ileum or neoterminal ileum, to predict future disease activity or assess the efficacy of interventions, can be carried out with standard cup forceps without significant contamination by colonic microbiota.

The gastrointestinal tract of humans is inhabited by a complex microbial community that is made up of 100 trillion cells, 10 times the number of human cells in the body.1 There is wide variation in the specific composition of the microbiota of the gastrointestinal tract between individuals, but the microbial community appears to be relatively stable within an individual.2 The composition of the intestinal microbiota has been implicated in the recurrence of Crohn's disease, as antibiotics (metronidazole and ornidazole) delay the recurrence in the neoterminal ileum after surgical resection.3 4 The study of the intestinal microbiota has been revolutionised by the use of molecular-based methods, many of which target the 16S rRNA-encoding gene. These methods circumvent the main disadvantage of traditional culture-based approaches, which fail to grow and identify more than half of the species in the gut.5–7 Terminal restriction fragment length polymorphism (T-RFLP) is a molecular method that utilises restriction enzyme digestion to identify 16S sequence heterogeneity. Sequence heterogeneity is then used as a proxy for different bacteria. T-RFLP is suitable for rapid, low-cost analysis of samples from multiple subjects.8 This approach has been used to study the intestinal microbiota in many published studies.9–16

Microbiota studies have traditionally been conducted on human biopsy samples from subjects using a standard unsheathed biopsy forceps. Contamination with bacteria that are not from the biopsy site2 17–20 can occur when the forceps passes through the colonoscope channel, particularly if colonic contents have been suctioned through the channel during the procedure. Eckburg et al2 showed that the abundance of constituent bacterial species of the microbiota is patchy and that the distribution of these species is heterogeneous from the small bowel to the colon. It is clear that human cecal microbiota differs quantitatively and qualitatively from the fecal microbiota.2 21 Furthermore, the local gut-associated microbiota is more likely to be involved in the pathogenesis of inflammatory bowel disease than the fecal microbiota.19 20 Therefore, in order to study the role of the microbiota in particular locations (ie, the neoterminal ileum), it is necessary to obtain biopsies from the site itself, rather than assuming that the fecal microbiota is representative of proximal locations in the gastrointestinal tract. The aim of this study was to determine whether sterile sheathed forceps are needed to obtain ileal microbiota samples that are not contaminated with colonic bacteria.

Methods

We recruited 40 consecutive subjects from the University of Michigan Medical Procedures Center scheduled for a clinically indicated colonoscopy. Inclusion criteria were 18 years of age or older and the successful performance of a full colonoscopy with terminal ileal intubation after three or fewer attempts. Exclusion criteria were an inability or unwillingness to consent to the study, use of antibiotics in the past 3 months, inability to intubate the terminal ileum within 3 min and any medical contraindication for obtaining biopsies.

Making the sheathed forceps from Radial Jaw 4 forceps

We used the Radial Jaw 4 (RJ4) single use biopsy forceps (Boston Scientific, Natick, USA) and modified them by covering the entire length with fluorinated ethylene propylene tubing to form a sheath (Zeus 0000048562, Orangeburg, SC, USA). To seal this tubing we incorporated a wax plug at the tip of the forceps. The wax plug was composed of four parts turnip wax (International Waxes, Ltd, Titusville, PA, USA, which meets the US Food and Drug Administration requirements set forth in 21 CFR 172.886 for use in food articles) and one part petrolatum jelly (Vaseline) to obtain a plug compatible with sterilisation that was easily extruded when the terminal ileum was reached. After forceps assembly in our laboratory, these were sterilised in ethylene oxide at the University of Michigan Hospital sterile supply services facility. Successful sterilisation was indicated by a change of colour on a gas sterilisation test strip. The University of Michigan Biomedical Engineering Unit certified the sheathed forceps for use in human subjects.

Assessment of sheath resistance to contamination

To determine whether sterility of the sheathed forceps was maintained during passage through the colonoscope, a colonoscope contamination ex-vivo experiment was performed without a human subject. A colonoscope channel was deliberately contaminated by suctioning a slurry of human stool through the colonoscope before insertion of either unsheathed or sheathed forceps. Each type of forceps was passed through the colonoscope and extended out of the colonoscope (and sheath, if present). The forceps were opened and closed, then withdrawn into the colonoscopy (and sheath, if present). The forceps were withdrawn from the colonoscope, extended and opened so that the jaws could be shaken in 1 ml of sterile saline. Bacterial contamination of the saline solution was assayed by testing for bacterial 16S ribosomal RNA by PCR using a broad-range forward primer 8F (Integrated DNA Technologies 5′-AGAGTTTGATCCTGGCTCAG-3′, Coralville, IA, USA) and a broad-range reverse primer 1525R (Integrated DNA Technologies 5′-AGA AAG GAG GTG ATC CAG CC-3′). PCR products were visualised on a 0.8% agarose gel stained with ethidium bromide and digitally photographed.

Biopsy sample collection from human terminal ileum

Before collecting each biopsy sample with a given forceps, the cecum was irrigated with water to produce a soup of stool, and one-quarter of the liquid contents were suctioned through the scope to simulate the worst-case scenario of significant stool contamination of the colonoscope channel. Subjects were randomly assigned to which forceps (sheathed vs unsheathed) were used first in a 1:1 allocation ratio. The allocation sequence was generated with the uniform function in Stata 10.1 by PDRH, and allocations were written on paper and sealed in opaque security envelopes until the patient was consented and sedated for colonoscopy. The participants were consented, and envelopes opened and randomisation order assigned by MNC and JF. We obtained two pairs of biopsy specimens from adjacent locations (proximal vs distal) in the terminal ileum using the sheathed and unsheathed forceps. The biopsy specimens were collected from a single pass with a new sterile (sheathed or unsheathed) forceps used for each biopsy specimen. Each biopsy specimen was transferred from the spike of the forceps using a sterile 20 gauge blunt needle to a sterile 2 ml screw-cap cryovial kept on dry ice. The tissue samples were immediately snap frozen in liquid nitrogen (−70°C) and transferred to the laboratory. The participants and the analysts of T-RFLP results (LAJ, STW, VBY, SW, VAY) were blinded to the type of forceps used for each sample. The study coordinators (MNC, JF) and the endoscopist (PDRH) could not be blinded to the forceps type.

Terminal restriction fragment length polymorphism

The microbiota of the terminal ileum was characterised for each sample using T-RFLP. This technique generates a community fingerprint based on the composition of the constituent bacterial members.12 22 Total DNA was extracted from biopsy samples as follows: The entire biopsy sample (20–50 mg) was transferred to a Mo Bio Ultra Clean Fecal DNA tube (Mo Bio Laboratories Inc, Carlsbad, California, USA) containing 350 μl of ATL buffer (Qiagen DNAeasy extraction kit; Qiagen). Tissue was disrupted in ATL buffer by homogenisation for 1 min using a BioSpec mini-beadbeater (BioSpec Products, Bartlesville, Oklahoma, USA). After disruption, samples were digested with 40 μl of 20 mg/ml proteinase K (Qiagen) at 55°C for 1 h. Total DNA was extracted as per the manufacturer's instructions using the Qiagen DNAeasy extraction kit (Qiagen) and eluted with 30 μl AE buffer. DNA were stored at −80° before PCR analysis.

The 16S rRNA encoding gene was amplified by PCR using a 6-FAM-5′-labelled, broad-range forward primer 6-FAM-8F (Integrated DNA technologies, Coralville, IA, USA, 5′-AGAGTTTGATCCTGGCTCAG-3′)23 and a conventional, broad-range reverse primer 1525R24 (Integrated DNA Technologies 5′-AGAAAGGAGGTGATCCAGCC-3′). PCR was performed with 70 ng DNA using illustra PuRe Taq Ready-To-Go PCR beads (GE Healthcare, Piscataway, New Jersey, USA). Cycling conditions consisted of an initial denaturing step at 94°C for 2 min followed by 30 cycles of 94°C 30 s/58°C 45 s/72°C 90 s and a final 4 min extension at 72°C. With every PCR run, a blank (no DNA template control) was included. Amplification was confirmed by visualisation of a single 1.4 kb PCR product on a 0.8% agarose gel. Biopsy specimens were considered adequate if 16S rRNA-encoding gene PCR products could be amplified from all four biopsy specimens from the same patient.

Amplicons were column purified (QIAquick PCR Purification Kit; Qiagen, Inc, Valencia, CA, USA) as per the manufacturer's instructions. Amplicon DNA (200–300 ng, as determined by Nanodrop spectrophotometer measurement (Thermo Scientific, Wilmington, Delaware, USA) was digested with 20 units of the restriction enzyme MspI (New England Biolabs, Ipswich, MA, USA) for 2 h at 37°C.25 Restriction fragments were column purified with the QIAquick PCR kit according to the manufacturer's instructions and eluted with 30 μl EB buffer. DNA concentration was determined using the Nanodrop spectrophotometer.

Fragments were separated on an automated capillary sequencer (Applied Biosystems 3730XL DNA Analyser); 100 ng Msp I-digested amplicons were loaded in duplicate onto a CE plate containing ROX1000 size standard. As a control, 100 ng undigested amplicons were run in duplicate on the same plate. Fluorescently labelled terminal fragments generated chromatogram peaks and were identified using Peak Scanner Software (Applied Biosystems). Peaks corresponding to fragments between 50 and 1000 base pairs (bp) in length were used in the analysis (reproducibility of peaks inside this range is high) and the height of each peak was obtained and used as a proxy for fragment abundance.

Visualisation and statistical analysis of community similarity

For community comparisons, a count matrix was generated, in which terminal fragment lengths (50–1000 bp) were rows, biopsy samples were columns, and each element in the matrix was the height of the given peak in a biopsy sample. The matrix was loaded into the EstimateS program26 and the similarity between all communities was calculated using the Bray–Curtis similarity index (BCI). BCI is an ecological diversity index that ranges in value between 1 (communities are identical) and 0 (communities are completely different) and takes into account the presence/absence of a species as well as its relative abundance.27 Because the results of T-RFLP identify distinct terminal restriction fragments (TRF), rather than species, ecological diversity measures were adapted to use TRF by using peak height as a proxy for TRF abundance.12 22 To visualise the similarity between all communities, a neighbour-joining dendrogram was constructed based on all pairwise BCI values using the MEGA4 program.28

As four samples (two sheathed and two unsheathed) were collected from each patient, we tested for differences in the average BCI between samples within patients and between patients by forceps type. Analysis of variance was used to test for statistical differences between these groups (SAS statistical software). If significant sample contamination occurred with the standard forceps, a sizeable increase in microbial diversity should be detected compared with unsheathed samples.

TRF richness (S) was calculated as the total number of individual T-RFLP peaks present in each biopsy sample. We also calculated the Shannon–Wiener diversity index (H′), which is maximised when there are many species (TRF) present in equal proportions.12 This was calculated based on the number of TRF (S) and the relative abundance of each TRF (pi), as follows:H=i=1spilnpiwhere pi is the proportion of the ith peak relative to the sum of all peak heights.

The sample size determination was based on previous analyses of intestinal microbial data from our microbiome core. The SD in the Shannon diversity index was expected to be less than 0.5 for paired samples. We prespecified a significant difference in the Shannon diversity index as a change of at least 0.5. With a power of 95%, and a two-sided α of 0.05, this would require a minimum of 13 subjects per group. Given that these were estimates, we increased the sample size to 16 for a margin of safety. We also allowed for four cases of dropouts or sample processing problems, increasing our planned sample size to 20 per group, or 40 overall.

This study was approved by the University of Michigan Institutional Review Board (IRB-MED) on 17 January 2008. The protocol for this study is posted for public review at: http://www.med.umich.edu/higginslab/protocols.html.

Results and discussion

Sheath sterility

Two pairs of unsheathed or sheathed forceps were passed through a deliberately contaminated colonoscope and analysed by PCR for the bacterial 16S rRNA gene. Both unsheathed samples were contaminated with bacteria, while the sheathed samples did not have detectable bacterial DNA (data not shown).

Patient samples

We obtained complete T-RFLP profiles from all four biopsy sites for 108 samples from 27 subjects. Of the 40 consented subjects, the first three subjects served as a pilot sample in which the biopsy specimen collection protocol was refined, but no microbiota analyses were done on these samples as poor quality DNA was obtained. In another six subjects, biopsies were not attempted due to endoscopy scheduling pressures and/or because ileal intubation could not be achieved in 3 min or less, as agreed to with the Institutional Review Board. In another four subjects, at least one of the four biopsy specimens did not yield any T-RFLP peaks, so they were excluded from the analysis. Despite these dropouts, randomisation by forceps type was successful and samples were collected using sheathed forceps first in 15 patients and unsheathed first in 12 patients. The baseline demographics of these subjects and their indications for colonoscopy are presented in supplementary table 1 (available online only).

T-RFLP analysis

It is important to note that T-RFLP uses broad-range nucleotide primers to amplify the same genetic locus (the 16S rRNA encoding gene) from all bacterial chromosomes represented in bulk extracted (community) DNA. Therefore, PCR amplicons represent all members of the bacterial community of a sample. As the forward primer in the T-RFLP PCR is fluorescently labelled and an endonuclease is used to digest amplicons at unique restriction sites, the length of TRF represents unique bacterial members of the community (ie, different alleles of the 16S locus vary in restriction sites) and can be quantified using a capillary nucleotide sequencer.25 Amplicons of the same size increase the fluorescent signal (chromatogram peak height) so that the area within a chromatogram peak is a proxy for the abundance of a particular bacterium. Therefore, data generated by T-RFLP are both the length of a terminal fragment (presence of a particular bacterium) and the abundance of a terminal fragment (the abundance of a particular bacterium). One community ecology metric (a standard of measurement) that incorporates both the presence/absence of a species and its relative abundance is called the BCI. This metric was used to evaluate the similarity between the communities (samples) in this study. To visualise the similarity among the communities, a dendrogram was constructed based on all pairwise (sample-to-sample) Bray–Curtis values (figure 1). To appreciate the similarity of two particular communities, one simply needs to note the branch length that connects each community. Shorter branch lengths represent similar communities, while longer branch lengths represent dissimilar communities.

Figure 1

Bacterial Community Diversity Dendrogram. This neighbour-joining dendrogram was based on all pairwise comparisons between samples taken with sheathed (n=2) and unsheathed (n=2) forceps from 27 patients. Each sample is represented by a branch in the tree and branch lengths (the length of the line that connects two tips) represent the similarity among bacterial communities in these samples (the shorter the length, the more similar the communities).

For the majority of patients, the tissue-associated microbiota obtained by both types of forceps (sheathed vs unsheathed) in a given patient were similar, as illustrated by the T-RFLP chromatograms (figure 2A,B), yet distinct from the microbiota of other patients, as illustrated by the dendrogram (figure 1).With few exceptions, samples obtained from an individual patient clustered together, indicating that samples from a single patient were more similar to each other than to samples from other patients. Also, no obvious clustering of sheathed and unsheathed samples was observed, indicating that the same bacterial community was sampled using the sheathed and unsheathed forceps.

Figure 2

Representative terminal restriction fragment length polymorphism (T-RFLP) chromatograms from unsheathed versus sheathed forceps. Representative T-RFLP profiles of unsheathed (A) compared to sheathed (B) biopsy samples from three patients. T-RFLP fragments (in base pairs) are represented as individual peaks. The peak height (relative fluorescence) corresponds to the relative abundance of each T-RFLP fragment. The most abundant individual fragments are labelled with respect to fragment size (in base pair) and relative percentage (decimal).

The difference in the microbiota between patients was much greater than the variability within patients, as has been observed in previously published studies.2 To determine whether sheathed and unsheathed forceps sample the same microbiota within patients, we tested for significant differences in diversity, as measured by the BCI. There was no significant difference in the microbial diversity for sheathed samples versus unsheathed samples (figure 3). We did multiple comparisons of the diversity of the microbiota, however in all groups the BCI measures overlapped. On average, the same microbiota was obtained from biopsies taken with sheathed and unsheathed forceps.

Figure 3

Microbial diversity measured by Bray-Curtis index (BCI). The mean difference in BCI between samples (pairwise analysis within individual patients) is presented. Error bars represent the SE of the mean. No significant differences by location or forceps type were found.

The mean TRF richness was virtually identical in the unsheathed versus sheathed samples, (602.57 vs 601.50; t-test p=0.803) (figure 4). A similar result was obtained using the Shannon–Wiener index (H'). On average, H′ was similar between forceps samples (7.51 for unsheathed vs 7.38 for sheathed) and was not statistically different (t-test p=0.316) (figure 5). The overall microbial diversity was thus not significantly different between samples obtained with the sheathed and unsheathed forceps. The range of H′ values calculated for sheathed samples was wider than that obtained for unsheathed forceps. This observation suggests that samples obtained using unsheathed forceps have less variability in TRF richness (TRF presence) or evenness (relative abundance of organisms). It is difficult to tell from these data the source of the variation in sheathed samples (richness or evenness), but it is clear that sheathed forceps do not consistently sample less or more bacterial diversity than unsheathed forceps. It is important to point out that these results are only as relevant as the individuals from which they were obtained. Given the uniqueness of the human gastrointestinal tract microbiota, it is possible that the mean and range of microbial diversity reported here (H′ values) may change as more individuals are sampled, but on average we find little evidence that these changes will be influenced by whether the samples are taken with sheathed versus unsheathed forceps.

Figure 4

Terminal restriction fragment (TRF) richness (S). The middle line in the box represents the mean TRF richness. There is overlap between the TRF richness of unsheathed and sheathed forceps with the mean TRF richness of unsheathed forceps being only slightly greater than that of sheathed forceps (t-test, p=0.803).

Figure 5

Terminal restriction fragment (TRF) diversity as measured by the Shannon-Wiener diversity index (H'). The middle line represents the mean Shannon-Wiener diversity index for each group. The Shannon-Wiener diversity index for each individual patient is represented by a filled square (unsheathed) or filled circle (sheathed). The H' overlaps between unsheathed and sheathed forceps with the mean H' of unsheathed forceps being only slightly more than that of sheathed forceps. This is not statistically significant (t-test, p=0.316).

The main limitations to this study are those related to the sensitivity of the PCR amplification and those related to the detection of PCR fragments. T-RFLP relies on PCR amplification with broad-range primers, which do not amplify the 16S sequences of some rare bacteria. In addition, two distinct species of bacteria can potentially produce the same T-RFLP peak in their profile, potentially obscuring some species diversity. Also, the ability of automated capillary sequencers to size fractionate fluorescently labelled PCR amplicons is dependent on the abundance of the amplicons, so organisms at low abundance are not represented in the analysis. More sensitive techniques, such as next-generation DNA sequencing, can identify low abundance species in the gut microbiota.29 The most important difference between T-RFLP and these newer techniques is one of scale. Whereas T-RFLP data represent the most abundant microbial TRF, next-generation sequencing can identify rarer members of the microbial community. So it is possible that characterisations using newer techniques would show real differences between samples collected with sheathed versus unsheathed forceps, but results obtained from T-RFLP suggest that the diversity among the dominant gut microbes will be similar.

In this worst-case scenario test, with deliberate and substantial contamination of the colonoscope channel with cecal contents, no significant differences were found in TRF diversity or TRF richness. This may be due to the use of forceps with a biopsy spike that is covered by an outer cup. Contamination of forceps in the colonoscope channel may only occur on the outside of the forceps cup, and it is possible that biopsies are adequately protected from bacterial contamination by the contents of the colonoscope channel once they are enclosed in the forceps cup.

Conclusion

Based on T-RFLP data, sheathed forceps do not appear to be required for microbiota sample collection from the terminal ileum. There are many factors that affect the composition of gut microbiota such as diet, recent use of antibiotics, concomitant intestinal disease and host genotype, which make it difficult to compare the microbiota between different individuals.30 31 The main strength of our study is that comparisons were made between paired biopsies in individual patients, thus eliminating the need to account for the above factors. While more sensitivity for rare species could be obtained with other techniques, this worst-case contamination study did not find significant differences. Based on these data, we would recommend the following for obtaining biopsies for microbial analysis. First, taking usual care to minimise contamination of the colonoscope channel, or flushing with sterile saline before passing the forceps; and the use of standard forceps in which the biopsy is retained with a biopsy spike and enclosed by a forceps cup during withdrawal of the forceps. We hope that this approach will accelerate the analysis of the microbiota of the neoterminal ileum in Crohn's disease to elucidate which bacteria contribute to postoperative recurrence.

Take-home message

For study of the microbiota of the mucosal surface of the intestine (eg, to study the microbiota that cause Crohn's disease recurrence in the of the neoterminal ileum after surgical resection), standard biopsy forceps can be considered sufficient for microbiota sampling.

References

Supplementary materials

Footnotes

  • Funding This study was funded by an endoscopic research award from the American Society of Gastrointestinal Endoscopy. The funding body had no role in the study analysis or writing of the manuscript.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was approved by the University of Michigan Institutional Review Board.

  • Provenance and peer review Not commissioned; externally peer reviewed.