Article Text

Download PDFPDF

Impression membrane for the diagnosis of microbial keratitis
  1. Stephen Kaye1,2,
  2. Henri Sueke1,2,
  3. Vito Romano1,
  4. Jern Yee Chen1,
  5. Nicole Carnt3,
  6. Stephen Tuft3,
  7. Timothy Neal4
  1. 1St Pauls Eye Unit, Royal Liverpool University Hospital, Liverpool, UK
  2. 2Department of Eye and Vision Science, Institute of Ageing and Chronic Disease, University of Liverpool, Liverpool, UK
  3. 3Corneal Service, Moorfields Eye Hospital, London, UK
  4. 4Department of Medical Microbiology, Royal Liverpool University Hospital, Liverpool, UK
  1. Correspondence to Professor Stephen Kaye, St Pauls Eye Unit, Royal Liverpool University Hospital, Prescot Street, Liverpool L7 8XP, UK; s.b.kaye{at}


Purpose To evaluate a corneal impression membrane (CIM) for isolation of bacteria, fungi and acanthamoeba in suspected microbial keratitis.

Methods Consecutive patients presenting with suspected microbial keratitis were included. For each patient, samples were collected in a random order using a surgical blade and a 4-mm-diameter polytetrafluoroethylene CIM disc, and transported in brain heart infusion broth. Risk factors, best corrected visual acuity (BCVA), size, location, depth and healing time of the ulcer were recorded. The microbial isolation rate was used to compare sampling methods.

Results 130 patients were included (mean age 62.6 years, SD 19.0). An antimicrobial had been used prior to presentation in 36 (27.7%) patients. Mean major and minor ulcer diameters were 2.1 mm (SD 2.0) and 1.6 mm (SD 1.7). Mean healing time was 12.4 days (SD 13.6). BCVA at presentation and following healing was 0.7 (SD 0.7) and 0.62 (SD 0.7) (p=0.34). There were 66 isolates (50.8%); 53 (40.8%) using a CIM and 35 (26.9%) using a blade (p=0.02). Staphylococcus aureus and coagulase-negative staphylococci were the commonest isolates. Isolation rate was not influenced by organism type, although in four cases Acanthamoeba spp. were isolated; three using CIM and one a blade.

Conclusions In this study, the isolation of microorganisms from cases of suspected microbial keratitis was significantly higher using a CIM than a surgical blade. A CIM may be a useful alternative or addition for sample collection in microbial keratitis.

  • Cornea
  • Diagnostic tests/Investigation
  • Infection
  • Microbiology
View Full Text

Statistics from


Although there are characteristic clinical features of microbial keratitis, identification and isolation of the causative microorganism is necessary to confirm infection and optimise clinical management,1 ,2 particularly because increased antimicrobial resistance3 ,4 has reduced the role of empirical treatment.5 While the PCR has shown promise as a diagnostic tool for the diagnosis of microbial keratitis,6 isolation of the organism is at present still needed to determine susceptibility to antimicrobials and other microbiological characteristics. Identification of the pathogen when there may only be a few organisms in a corneal ulcer requires an adequate sample cultured on a variety of different media.2 ,5 ,7 ,8 It is usual to collect multiple scrapes from the ulcer with a disposable needle or surgical blade, or with a reusable spatula and to plate the material onto several culture media.2 ,5 ,7–9 Collecting multiple samples with a sharp instrument, particularly from an uncooperative patient or a pending perforation, however, may be difficult and requires skill and biomicroscopy. Keeping fresh culture materials in the clinical area may also be logistically difficult, and a variety of techniques have been used to simplify the isolation process.8 ,10 For example, placing a sample directly into transport media such as brain heart infusion (BHI) broth8 ,10 or Amies transport medium11 before plating in the laboratory gives comparable results to direct plating in clinic with no loss of sensitivity or specificity, including the isolation of fungi, and enables sub-culturing on enrichment media for fastidious organisms.8 ,10

The collection of ocular surface cells by impression cytology using cellulose acetate filter paper was first described in 1977.12 Cellulose impression paper has more recently been used in the diagnosis of viral, fungal and acanthamoeba infections.13 ,14 Hydrophilic polytetrafluoroethylene (PTFE) impression membranes have also been used to diagnose ocular surface neoplasia and viral infections.15 ,16 Impression membranes have also been used in combination with flow cytometry to phenotype leucocytes and inflammatory markers on the ocular surface in normal, ageing and various ocular surface diseases.17 For the diagnosis of microbial keratitis, an impression membrane has the potential advantage of covering a greater surface area of a corneal ulcer and, being less invasive than a blade or needle, it may be safer and more suitable for general use. To evaluate whether the use of an impression membrane is as sensitive as a blade, we compared the isolation rate using the two techniques in a series of patients who presented with suspected microbial keratitis.


We included consecutive patients with suspected bacterial, fungal or acanthamoeba microbial keratitis seen between October 2012 and December 2013. Patients with a suspected herpetic keratitis were excluded. The corneal impression membrane (CIM) discs were prepared from a sheet of hydrophilic membrane PTFE (Biopore Membrane, Merck Milipore, UK) using a 4 mm punch and autoclaved at 136°C for 45 min. For each patient, samples were collected using a surgical blade and a CIM, in a random order, as follows. Wearing sterile gloves, a number 11 blade (Swann Morton, Sheffield, UK) was used to scrape the edges of the ulcer and the blade placed in a bottle containing 0.5 mL of BHI broth as previously described.8 A CIM was placed over the ulcer using sterile single use forceps or, if the ulcer was >4 mm in diameter, the CIM was placed to include the edge of the ulcer. The eyelids were kept apart and care was taken to avoid contact with the eyelids or conjunctiva. The membrane was left in position for approximately 5 s with the eye held open and then removed and placed in 0.5 mL BHI broth. A second scrape was then taken using an 11 blade for a smear for Gram stain.

BHI bottles and slide smear were then transported to the microbiology laboratory at room temperature. On receipt, unless at night, the bottles were vortexed for 5–10 s and 10 µL aliquots inoculated onto blood, chocolate, fastidious anaerobic and Sabouraud's dextrose agar plates and Robertson's cooked meat enrichment broth (Oxoid, Basingstoke, UK).8 ,11 ,18 ,19 Finally, a non-nutrient agar (Oxoid) seeded with Escherichia coli was inoculated with 20 µL of broth for Acanthamoeba spp. culture.11 If the samples were received at night, the bottles of BHI were kept at room temperature as this facilitates growth of the microorganisms within the medium.8 ,11 ,18 ,19 Agar plates and a 24 h subculture of the BHI broth in enrichment medium were examined for evidence of bacterial growth after 24 and 48 h incubation. All isolates were identified using matrix-assisted laser desorption ionisation-time-of-flight mass spectrometry (Bruker, Bremen, Germany). Acanthamoeba plates were examined daily for 7 days for the characteristic appearances of acanthamoeba trophozoites and cysts.

Patients were initially treated with a topical fluoroquinolone with additional treatment according to the clinical picture at presentation, the microorganism isolated and susceptibility data. Patients were examined daily for 3 days and then twice a week until healing, defined as time in days to re-epithelialisation of the ulcer. The following clinical information was collected: type of treatment that the patients had received or were still using at presentation, risk factors such as contact lens wear, presence of ocular surface disease, previous corneal surgery and previous microbial keratitis. Best corrected visual acuity (BCVA), size, location and depth of the ulcer and healing time were recorded. BCVA was converted into logarithm of the minimum angle of resolution units. Ulcer size (minor and major axes) and location (minimum distance from the limbus) were measured to the nearest 0.01 mm.20 The ulcer depth was graded on a nominal scale of 1–4 based on the proportion of corneal thickness lost on slit-lamp biomicroscopy (0<1<25%, 25%≤2<50%, 50%≤3<75% and 75%≤4<100%). Fisher's exact and Wilcoxon tests (SPSS V.22) was used for analysis, and p<0.05 was considered as significant.


In total, 130 consecutive patients (69 males and 61 females) were included (mean age 62.6 years, SD 19.0). Twenty-six (20%) patients had worn contact lenses, 17 (13.1%) had a previous history of microbial keratitis and 10 (7.7%) had previous corneal surgery. Additional possible risk factors in the remaining patients included blepharitis, acne rosacea, trauma and atopy. A topical antimicrobial had been used immediately prior to presentation in 36 (27.7%). The mean ulcer depth was grade 1.2 (median 1.0, SD0.4). Ulcers were, on average, located 2.2 mm (median 2.0 mm, SD 1.2) from the limbus with major and minor diameters of 2.1 mm (SD 2.0) and 1.6 mm (SD 1.7), respectively. The mean healing time was 12.4 days (median 10 days, SD 13.6). There were no perforations or loss of an eye and no patients required surgical intervention (corneal transplant, amniotic membrane graft or corneal gluing). There was no difference in healing time between patients who had (12.8 days, SD 12.8) or did not have a microbial isolate from their ulcer (12.5 days, SD 11.8) (p>0.5). BCVA at presentation and following healing was 0.7 (SD 0.7) and 0.6 (SD 0.7) (p=0.34).

The overall isolation rate (CIM and/or blade) was 50.8% (66 isolates from 130 participants). The isolation rate using a CIM was 40.8% (53/130) and a blade 26.9% (35/130) (p=0.02). No organisms were detected on Gram stain. In 16.9% (22/130) of cases, an isolate was identified from both the CIM and the blade. In 31 cases (24%), a microorganism was isolated from the CIM but not the blade and, conversely, in 12 cases (9%) a microorganism was isolated from the blade but not the CIM. There were no significant differences in ulcer area, depth or distance from the limbus between ulcers with positive or negative cultures in any group (p>0.5) (table 1).

Table 1

Clinical characteristics of ulcers from 130 cases of microbial keratitis related to method of culture and isolation of microorganism

The use of an antimicrobial prior to presentation did not significantly affect the isolation rate, whether using a CIM or a blade. With a CIM, the isolation rate for patients who had or had not received an antimicrobial was 16 of 36 (44%) and 37 of 94 (39%) (p=0.69) and with a blade it was 8 of 36 (22.2%) and 27 of 94 (28.7%) (p=0.51).

The microbial isolates are shown in table 2. As a percentage of the microorganisms isolated, Staphylococcus aureus and coagulase-negative staphylococci (CNS) were the commonest bacteria isolated. Of the four cases where there was a mixed infection, a CNS was isolated with either a Moraxella nonliquefaciens, a Bacillus sp. or an α-haemolytic streptococcus; and in one case, an Acanthamoeba spp. and an α-haemolytic streptococcus were isolated. In two (2%) cases, a different microorganism was isolated from the blade and the CIM (blade: CNS, CIM: Moraxella sp., blade: α-haemolytic streptococcus, CIM: Bacillus sp. and CNS). There were no significant differences between a blade and CIM in terms of the type of microorganism isolated; S. aureus (p=0.26), CNS (p=0.11) and Streptococcus spp. (p=0.10). In four cases, Acanthamoeba spp. were isolated, three using CIM and one a blade (p=0.62).

Table 2

Isolates from cases of presumed microbial keratitis grouped according to sampling method: blade or corneal impression membrane (CIM)


Identification of the causative microorganism and susceptibility profile forms an important component in the management of microbial keratitis.7 Recent attempts to improve the diagnostic yield from the investigation of microbial keratitis have focused on PCR to amplify microbial DNA. For bacteria, this involves either the amplification of a conserved region of DNA from the gene encoding the bacterial ribosomal RNA and then sequencing the product to identify the type of bacteria or using a set of primers specific to known bacterial pathogens. Acanthamoeba and fungi can also be detected by PCR.21 ,22 Larger comparative studies, however, are necessary to fully ascertain the sensitivity and specificity of PCR compared to culture for the diagnosis of microbial keratitis.

Culture of the causative microorganism is necessary to determine the susceptibility profile of a pathogen, which is still an important component in the management of microbial keratitis.2 ,7 Choosing a safe and effective method to isolate and grow the pathogen is therefore of prime importance. Our results indicate that the isolation rate using CIM (40.8%) was significantly higher than we achieved using a surgical blade and comparable with other previous published reports. Levey et al18 used a Kimura spatula and reported a positive culture from 47.1% of 119 cases, while Epley et al19 used a mini-tip Culturette swab and obtained a positive culture from 42% from 12 cases. Further comparisons between a CIM and other sampling methods such as a Kimura spatula or alginate swab would be important.

In our series, treatment with a topical antimicrobial immediately prior to culture did not significantly reduce the isolation rate. This is in contrast to a previous report using a blade in which the isolation rate was 27% from cases on treatment compared with 42% off treatment.8 The rate using a CIM sampling alone was (41%), which is similar to the report of Morlet et al,23 who achieved an isolation rate with needle sampling and direct plating of 41.5% from 118 patients, of whom a similar percentage (22%) had received a topical antimicrobial prior to investigation.23 Although use of a blade or a needle11 allows material to be obtained from the leading edge of an ulcer and may help debridement, a CIM offers the advantage of sampling from a greater surface area of an ulcer. It is also simple to use and could lend itself to nurse-led sampling. Whether isolation rates would be increased by taking multiple CIM samples is unknown, but given the simplicity and ease of collecting samples it is worthy of further investigation. It is of note that if we had taken only one sample obtained by CIM, 9% of isolates would not have been detected, whereas if only one sample had been obtained by blade 24% of isolates would not have been detected.

The microbiological spectrum of corneal isolates from this series is similar to a recent report from the UK,24 apart from a higher detection of S. aureus and a lower isolation of CNS (28% and 23%). The lower isolation rate of CNS may be important as the role of CNS as a pathogen for suspected microbial keratitis is unclear and difficult to distinguish from a contaminant.20 In terms of the other microorganisms isolated, there were no differences in the isolation rate between the CIM and blade except for the detection of Acanthamoeba spp., which in three cases was only isolated by a CIM. This agrees with a previous case series of isolation of Acanthamoeba spp. with impression cytology using fixation and light microscopy.14

It has previously been shown that the use of suitable transport media, such as Amies transport medium or BHI, followed by use of an enrichment medium gives isolation rates comparable to direct plating on culture media, including for the isolation of fungi, with no loss of sensitivity or specificity.8 ,10 ,11 BHI provides excellent recovery for a range of fastidious organisms and is an established medium for holding stock cultures in the laboratory at room temperature.18 ,19 Samples can, therefore, be transported to the laboratory in BHI or Amies medium at room temperature and kept overnight.8 ,11 There is also evidence that BHI directly supports the growth of a range of anaerobes, which can be further enhanced by the addition of 0.1% agar to the medium.25 Use of such transport media has certain advantages and enables sub-culturing on enrichment media for fastidious organisms. It is important that transport media support the growth of fastidious organisms because there is otherwise a potential risk of undergrowth of these fastidious microbes in a polymicrobial infection.11 It is of note that we found a very low rate of mixed polymicrobial infection in this and previous studies.8 In the clinical situation, where there may be very few organisms available from a corneal scrape, processing for fastidious organisms may be best done in the laboratory by a technician trained in microbiological methods. The lower between-scrape variation using an indirect method may also improve the probability of inoculating the appropriate culture medium in the laboratory.8 Nevertheless, a comparison of direct plating and BHI using a CIM would be of value.

In this study, although we did not collect an additional CIM for a Gram, Giemsa or KOH stain for direct light microscopy, it has been reported that PTFE membranes such as the CIM have significant advantages compared with cellulose acetate membranes for conventional impression cytology. This may be important for the diagnosis of fungal keratitis7 ,13 because the CIM membrane is transparent when wet or when treated for oil immersion oil transmission microscopy.16 In addition, a PTFE CIM membrane has a very low background fluorescence, which allows the use of fluoresceinated antibodies.16 Although we have shown a CIM to be a useful alternative in the clinical situation, further in vitro work and or an animal model would be helpful to determine thresholds for detection of different microorganisms (including herpes simplex virus 1) whether by culture or PCR.

In conclusion, the collection of corneal samples using a CIM that is then transported to the laboratory is an effective and safe system for general use. In our experience, the PTFE CIM has good mechanical stability and was easy to handle. A CIM provides a useful alternative or addition to the collection of samples from patients with suspected microbial keratitis.


View Abstract


  • Deceased May 2015.

  • Contributors SK, HS, ST and TN designed the study. SK was the principal investigator. JYC, NC, HS, VR and SK undertook the sample and data collection, TN and HS the microbiology, SK, HS, ST and TN the data analysis and drafting of the manuscript. All authors helped interpret data and critically revise successive drafts of the manuscript.

  • Funding ST is supported by the National Institute for Health Research (NIHR) Biomedical Research Centre. NC is funded by the Australian Government National Health and Medical Research Council (NHMRC).

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval National Research Ethics Service (NRES Committee North West, UK. The described research adhered to the tenets of the Declaration of Helsinki.

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

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.

Linked Articles

  • At a glance
    Keith Barton James Chodosh Jost Jonas