Point-of-Care Autofluorescence Imaging for Real-Time Sampling and Treatment Guidance of Bioburden in Chronic Wounds: First-in-Human Results

Background Traditionally, chronic wound infection is diagnosed by visual inspection under white light and microbiological sampling, which are subjective and suboptimal, respectively, thereby delaying diagnosis and treatment. To address this, we developed a novel handheld, fluorescence imaging device (PRODIGI) that enables non-contact, real-time, high-resolution visualization and differentiation of key pathogenic bacteria through their endogenous autofluorescence, as well as connective tissues in wounds. Methods and Findings This was a two-part Phase I, single center, non-randomized trial of chronic wound patients (male and female, ≥18 years; UHN REB #09-0015-A for part 1; UHN REB #12-5003 for part 2; clinicaltrials.gov Identifier: NCT01378728 for part 1 and NCT01651845 for part 2). Part 1 (28 patients; 54% diabetic foot ulcers, 46% non-diabetic wounds) established the feasibility of autofluorescence imaging to accurately guide wound sampling, validated against blinded, gold standard swab-based microbiology. Part 2 (12 patients; 83.3% diabetic foot ulcers, 16.7% non-diabetic wounds) established the feasibility of autofluorescence imaging to guide wound treatment and quantitatively assess treatment response. We showed that PRODIGI can be used to guide and improve microbiological sampling and debridement of wounds in situ, enabling diagnosis, treatment guidance and response assessment in patients with chronic wounds. PRODIGI is safe, easy to use and integrates into the clinical workflow. Clinically significant bacterial burden can be detected in seconds, quantitatively tracked over days-to-months and their biodistribution mapped within the wound bed, periphery, and other remote areas. Conclusions PRODIGI represents a technological advancement in wound sampling and treatment guidance for clinical wound care at the point-of-care. Trial Registration ClinicalTrials.gov NCT01651845; ClinicalTrials.gov NCT01378728


Introduction and Rationale
Wound care is a major clinical challenge and presents an enormous burden to health care worldwide[1 -3]. As wounds (chronic and acute) heal, a number of key biological changes occur at the wound site at the tissue and cellular level [2]. Among these are inflammation, reformation of the epidermal barrier, and remodeling of the connective tissue in the dermis.
However, a common major complication arising during the wound healing process, which can range from days to months, is bacterial infection [2,3]. This can result in a serious impediment to the healing process and lead to significant complications, especially in chronic non-healing wounds. Currently, the standard wound care includes monitoring for possible infection by direct visual inspection under white light and by taking samples for analysis in the laboratory which takes approximately two days to provide a result. However, qualitative visual assessment only provides a gross view of the wound site (i.e., presence of purulent material and crusting) but does not provide the critically important information about underlying changes that are occurring at the tissue and cellular level (i.e., infection, matrix remodeling, inflammation, and necrosis).
All chronic wounds contain bacteria, but identifying whether the wound is in bacterial balance (e.g. contamination with organisms on the surface or colonization with organisms in the tissue arranged in micro colonies without causing damage) or bacterial imbalance (e.g. critical colonization and infection) is of primary clinical importance as it is a major determinant of the rate of healing. Also, it is important to note that there is a continuum of bacterial presence progressing from bacterial balance to imbalance leading to bacteria-mediated tissue damage in a chronic wound. These bacteria include common species typically found at wound sites (i.e., Staphylococcus and Pseudomonas species) [1, 2,10,11]. The diagnosis of infection is typically made clinically, based on standard signs and symptoms [3] identified during examination in and around the local wound bed, the deeper structures, and the surrounding skin by the wound care team. However, a major problem in conventional practice is that bacteria within and around a for bacterial presence is suboptimal, despite the collection of multiple swabs [15]. Furthermore, bacteriological culture results often take about 2-4 days to come back from the laboratory, thus significantly delaying diagnosis and treatment [15]. Thus, bacterial swabs do not provide realtime detection of infectious status of wounds. In addition, although wound swabbing appears to be straightforward, it can lead to inappropriate treatment, patient morbidity and increased hospital stays if not performed correctly. Therefore, an image-based method that allows realtime monitoring of wound healing, particularly early dermal connective tissue remodeling, and the presence of bacterial contamination and/or infection over time could have a significant clinical impact.
Another aspect of bacterial impact on wound healing relates to connective tissue breakdown. The remodeling and healing of connective tissues in wounds involves simultaneous synthesis and degradation of collagen fibrils [8,9]. This degradation process is driven by bacterial imbalance with the host which leads to increased enzymatic-breakdown by proteases released by the bacteria. Changes in wound connective tissue (e.g. collagen in the surrounding Autofluorescence imaging provides a powerful means of visualizing both bacteria and connective tissue in real-time. Autofluorescence has been used previously (by our own team) in other clinical applications, for example, in gastroenterology to image both collagen and bacterial fluorescence in clinical studies [5][6][7]. Thus, we wish to expand the use of tissue autofuorescence imaging technology to wound care and management in order to obtain biologically relevant information about the wound site at the tissue and biomolecular levels in real-time during the healing process [5][6][7]. When used to assess wounds, tissue autofluorescence may aid in determining the degree of wound healing (e.g. by visualizing and measuring connective tissues) and the presence of bacterial infection.
In preliminary preclinical testing, we have discovered that when wounds are illuminated by violet/blue light, endogenous collagen in the connective tissue matrix emit a characteristic green fluorescent signal [6], while most pathogenic bacterial species emit a unique red fluorescence signal due to the production of endogenous porphyrins [5]. Therefore, with autofluorescence imaging, no exogenous contrast agents are needed during imaging, making this approach particularly appealing as a diagnostic imaging method for clinical use.

Study Objectives and Specific Aims
The primary objective of this clinical study is to evaluate the use and effectiveness of the UHN handheld PRODIGI™ imaging device for real-time and non-invasive detection and tracking of pathogenic bacterial presence, contamination and infectious status in complex wounds over time. This will enable us to determine if PRODIGI™ can detect and longitudinally and quantitatively track intrinsic changes that may occur during the wound healing process including, but not limited to, collagen re-modeling and bacterial infection of the wound site.
As a secondary objective, based on findings to date from past clinical studies (UHN REB 09-0015-A), we now wish to determine if PRODIGI TM can detect the presence of clinicallysignificant (critical colonization or infection) using fluorescence and then that information may be used by the clinical team to effectively deliver a clinical intervention to treat the infection.
As a third objective, we wish to obtain valuable end-user data on the clinical utility of the device within the wound clinic environment with a brief question-based survey. This information will be used to optimize subsequent versions of PRODIGI™ during product development and to improve the integration of the technology from the perspective of both the clinician and the patient.
Most importantly, our aim is to quantitatively verify the added value that this imaging device brings to traditional wound care practice and to what extent it will change the gold standard (microbiological swabbing) of clinical wound diagnosis. compared with changes in wound size over time.

2)
To determine the effectiveness of the fluorescence imaging device in detecting the presence (or contamination) of bacteria in and around a wound (including infection), compared with standard best practice methods using white light visualization and clinical signs and symptoms, with swabbing and bacteriology as the 'gold standard'.

3)
To identify the relationships between autofluorescence imaging of collagen and bacteria in wounds (including the wound margin) and the following: i) clinical signs of infection, ii) microbial load (i.e., number of organisms per gram of wound tissue), and iii) diversity of microbial species in the wound (i.e., number of different species isolated per wound), and Gram signing.

4)
To assess the potential utility of the device to guide intervention and alter the course of treatment in wounds that are infected.

Hypothesis
We hypothesize that real-time imaging of tissue autofluorescence signals emanating from endogenous connective tissue (e.g. collagen) and pathogenic bacteria within complex wounds can be used to determine healing status (i.e., collagen re-modeling and wound closure), detect wound bacterial contamination and/or infection that is occult under standard clinical white light evaluation, and guide intervention during wound care. be treated by staff physicians at JDRTC will be considered for eligibility into study. Eighty new patients with wounds will be entered into this trial, based on statistical power-based calculations for sample size determined in collaboration with the UHN Biostatistics Group (assuming one wound per patient). Recruitment of patients will continue throughout the study duration, until the total number of patients has been met. Recruitment of patients will be directed by Drs. Linden and Fedorko. After the recruitment process, consent will be obtained by a member of the UHN research team. The consent process will be documented in the Informed Consent Form

Clinical Setting and Patient Enrollment Criteria
Checklist as per UHN Best Clinical Practice Guidelines.

Patients will be included in the study according to the following criteria:
 > 18 years of age  males and females  new patient to the JDRTC to ensure consistent work-up procedures (as described below) prior to treatment  presenting with acute or chronic wounds (i.e., diabetic ulcers or other), with known or unknown infection status.

Patients will be excluded in the study according to the following criteria:
 treatment with an investigational drug within 1 month before study enrolment  any contra-indication to routine wound care and/or monitoring

Specific Study Procedures Standard White Light-Based Clinical Wound Assessment
Patients who meet the entry criteria (determined by staff at the JDRTC) with chronic wound problems will be informed and consent will be obtained to participate in this trial by the Data regarding the type, location, and history of the study wound will be collected from direct observation, the patient record and patient/caregiver report by the clinical team. In order to measure wound depth, a cotton-tipped swab will be placed in the deepest area of the wound and marked at the point level with the surrounding peri-wound skin. For chronic diabetic wounds the type and amount of wound bed tissue will be measured using the necrotic tissue subscale of the Pressure Sore Status Tool (PSST) [12]. Necrotic tissue will be defined as yellow, tan, brown, gray or black tissue in the wound bed after cleansing the surface of the wound bed with saline-moistened gauze, which ensured the necrotic tissue will be adherent [13].  For subjects with more than one eligible chronic wound (i.e., full-thickness and nonarterial), one wound will be randomly selected (i.e., random draw) for data collection procedures. If other 'eligible wounds' are available for a single patient, imaging of this wound will be conducted at the discretion of the physician and based on the time available to the clinical team for the additional procedure. All study data will be collected by a member of the research team and will be recorded on a Case Report Form labeled with subject/patient identification number only, which will be kept in secure storage (double locked) location during the study at the JDRTC.
After the initial data is collected for each wound, a high-resolution white light photograph (using a length scale placed in the field of view to aid in wound size measurements) will be taken of each wound using the imaging device and printed on a color printer, immediately after the wound has been examined. The clinician (or wound care personnel) will be asked to draw (using an indelible ink pen) on the photograph to indicate the area he/she considers infected with bacteria and thus would normally take a bacterial swab or biopsy (each photo will be labeled with patient ID, date and wound ID). However, no swabs or biopsies will be taken at this time. Immediately after the physician has 'committed' to their decision of bacterial infection on the wound and areas that would require swabbing for microbiology confirmation at diagnosis, PRODIGI™ will subsequently be used to take fluorescence images of the wounds. images into its digital memory card for analysis afterwards. If bacteria are detected on the second pass (using the fluorescence imaging device and spectroscopy measurements where appropriate) which are undetected during the standard clinical wound assessment procedure (first pass), then bacterial swabbing and/or tissue biopsy will be obtained from these bacteriared and bacteria-green fluorescence positive areas and sent for bacteriology testing to a blinded microbiology lab. Conversely, if bacteria are not detected on the second pass using fluorescence imaging but were detected during the standard clinical wound assessment procedure (first pass), then bacterial swabbing and/or tissue biopsy will be obtained from these areas and also sent for bacteriology testing to a blinded microbiology lab.
Multiple point fluorescence spectroscopy measurements will be made, at the discretion of the research team, to determine specific autofluorescence spectral signatures of unique fluorescence features observed during imaging. This will be within and around the wound using a custom-built laptop computer-controlled portable 'spectroscopy' device (See Fig. 2

, Study
Instruments). If deemed appropriate, areas of the wound bed and edge will be measured using the spectroscopy probe. Fluorescence imaging and spectroscopy measurements require that the room lights be dimmed or turned off during the examination for a brief time, to minimize the background light.
This device uses an optical fiber probe, which is held close to the wound surface (but not in contact) to collect the fluorescence signal from the wound and sends this information to the spectrometer which separates the light signal into different colors (wavelengths), and this data is stored on the laptop computer. Tissue spectra will be collected using the same blue light LED (405 nm emission) of the fluorescence imaging device. This measurement will be made after the fluorescence digital imaging is completed for each wound, and is estimated to take about 1-2 minutes. Spectra will be collected from areas of interest as determined by the fluorescence imaging device. Any areas that appear abnormal under fluorescence (e.g. suspected application of a topical skin cream, lotion, etc.) will also be measured to create a database of possible confounding non-biological autofluorescence signals. Each area measured by spectroscopy will be documented on a digital white light image of the wound area for comparison with bacteriology results. The fluorescence spectroscopy data will yield quantitative spectral information as to the relative intensity and contribution of the green and red fluorescence to the total fluorescence signal measured at the wound. The fluorescence imaging and spectroscopy data obtained from these wound measurements will be stored on an encrypted laptop computer for subsequent analysis, which will be stored in a secure (doublelocked) location at the JDRTC.

Swabbing and Bacteriology
A swab or biopsy will be collected from each area marked as being suspicious for infection in the printed color photos of both the white light and autofluorescence images taken with the prototype device. The swab/biopsy will be done in an area which has consensus from at least two study staff in order to determine the bacterial burden in each marked area. The study staff will indicate consensus by both initialing the fluorescence image after the areas have been marked.
For each marked area (e.g. from the white light and fluorescence color photos), a bacterial swab or a specimen of viable wound tissue will be removed aseptically using a standard sterile swab or a dermal punch instrument, and this will be determined by the wound care specialist/physician. The tissue specimen will be placed in sterile container and transported to the microbiology laboratory for processing by the end of the day. These will be used for quantitative bacterial culture analyses within the wound bed and from surrounding normal tissue. Wound cultures will be processed at a single microbiology laboratory associated with the JDRTC. Wound cultures (swabbing/biopsy) will be used to identify bacterial species present in and around the wound, as well as the relative number of colony forming units (CFU) and the gram signing of the bacteria. All organisms isolated will be identified using standard microbiologic procedures, which are based on criteria such as colony morphology and gram stain appearance [15].
Therefore, for each patient in this study, a complete data file will include (for each visit to the clinic): white light visualization and clinical signs and symptoms of each wound examined (performed by the clinician), white light and corresponding autofluorescence images of the wound obtained using the prototype device, corresponding fluorescence spectra (performed by research staff when appropriate) and bacteriology results obtained from swabbing/biopsy (performed at bacteriology lab). These data will be collated into and stored in a global database for the duration of the study and this will be stored on an encrypted computer in a secure location at the JDRTC.

Question Based Survey
A semi-structured survey will be administered that combines both numeric and qualitative comment boxes, to assess the informational needs of the target population. The questions will be closed-ended to provide discrete numerical or nominal data points. The patient survey will generally include two broad sections: 1. Demographic data of the patients. These will specifically include age, level of education, and annual household income.

Handling of Research Equipment During Study
The prototype fluorescence imaging device and the spectroscopic optical fiber probe will not come in contact with patients during imaging and spectroscopy measurements. However, to further increase sterility of all research equipment between patients and over time, after each imaging and spectroscope session, all equipment surfaces (e.g. prototype device, printer, computer) will be wiped clean using 70% ethyl alcohol. To date, we have not had any safety or contamination issues with the PRODIGI™ device in parallel clinical testing.

Longitudinal Imaging of Each Patient
Patients enrolled in this study will be followed over a period of time since wound assessment is performed over several visits, lasting weeks to months. Patients visit the JDRTC for pre-intervention "work-up" and post-intervention follow up care according to a predetermined schedule determined by their clinician and wound care team. Therefore, this study will involve to quantitatively track changes in the PSST, SCCS, or BWAT over time. We recognize that many wounds may not close prior to the end of the study therefore, we will measure and track the changes in wound size (and depth) during all visits for a patient until the end of this study.
Furthermore, the number of pre-intervention "work-up" and post-intervention follow up visits may vary due to variability in wound etiologies, unpredictable efficacy of the treatment, and logistical difficulties in scheduling. The research staff will accommodate the fluctuations in scheduling as much as reasonably possible, however it is understood that the longitudinal assessment of the wounds may in some cases be inconsistent with respect to the number of follow-up visits. We do not anticipate that an inconsistent number of follow up visits will compromise the integrity of the data collected or its analyses/interpretation within the scope and goal of this study.
All imaging and spectral data obtained will be stored with a unique patient identification number which will facilitate consistent tracking of patients over several clinical visits. Table 1 shows an example of the various procedures that will be performed during each patient visit over the course of the study.

Data Collection
All study data (demographics, images, survey) will be collected by a member of the research team and will be recorded on a Case Report Form labeled with subject identification number. Demographic data will be collected from the patient record. For subjects with more than one eligible chronic wound, one wound will be randomly selected (i.e., random draw) for data collection procedures. Data regarding the type, location, and history of the study wound will be collected from direct observation, the patient record and patient/caregiver report. Data collected during this study may be used to affect patient care. The clinical staff will determine the appropriate intervention, understanding and considering the utility and limitations of the fluorescence image. The intervention will be documented by the research team.

Intervention
The physician will decide the course of treatment for the patient based on clinical best practice and gold standard of care prior to seeing the fluorescence images. The scientific personnel will record this decision in the Case Report Form. The physician will then view the fluorescent images and re-evaluate the course of treatment. The scientific personnel will record this decision and note any discrepancy to the original treatment plan, and record the reason that the fluorescence images did or did not influence the clinical judgment about altering the course of treatment. Intervention may include, but is not limited to saline wash, hyperbaric oxygen as verification) shows that a wound is either critically colonized or infected, and this was missed during standard white light examination of the wound, and this new information led to the timely use of a treatment/intervention that in turn decreased bacterial load thereafter (confirmed by three weeks of post-intervention monitoring), then we will consider this clinically significant.

Adverse events
There are no known or perceived risks to the patient associated with this study as compared to the usual standard of care. The fluorescence imaging device (which is CSA approved for this application) and the fluorescence spectroscopy device used to study wounds in this trial use a harmless blue light (no heat is emitted from the light sources or the device) to illuminate the surface of the wound and then collect a digital image and spectrum, respectively, of the resulting tissue autofluorescence from the wound site. No exogenous drugs/agents are involved. Imaging of wounds is performed in a darkened room in a 'non-contact' and noninvasive manner so that no device or probe is in contact with the wound surface. Wound images are collected in the same way that a digital camera takes a photo from a distance. The fluoresence imaging and spectroscopy procedures pose no known risks to the patient or attending medical staff.

Data Storage and Statistical Analysis
Data Storage: all white light and fluorescence imaging and spectral measurements as well as corresponding bacteriology data will be stored in a global database (on an encrypted PC/laptop) with metrics to quantitatively describe the changes in fluorescence (green and red signals) determined over time to correlate with changes in wound healing and bacterial contamination/infection status. The data will be securely (double lock) stored at the JDRTC with access granted to only staff listed on the current protocol.
Statistical Analysis: we will engage the Biostatistics group at UHN to perform statistical analysis of the study data. For this purpose, mixed statistical models will be employed with wound closure (based on wound size, shape, depth) and bacterial contamination and/or infection (based on the clinical signs and symptoms of infection) as outcomes and the fluorescence intensity (either total or region-of-interest) for green and red emission channels, bacteriology results (bacterial CFU, bacterial species present, and Gram signing) and the timepoint of imaging as explanatory variables.
A major aim of this trial using the prototype PRODIGI TM fluorescence device is to gather important qualitative information about its general utility within the clinical wound care environment. For example, we are interested in learning from clinicians and other wound care personnel about the ease of use of the device and how it will be best used in the conventional clinical setting. In addition, we wish to determine the value of various existing features and capabilities as well as potential additional features required in subsequent versions of the device. In addition to this qualitative evaluation, we will also conduct quantitative analyses of all imaged and measured data obtained during this study, using established statistical methods.
The ability of the prototype PRODIGI™ fluorescence device to image wound remodeling (i.e., changes in collagen concentration and biodistribution) during wound healing will be determined by correlating the increase in green (collagen) autofluorescence intensity and biodistribution within the wound bed and at the wound boundary with measurements of the wound size (e.g. white light grid-based tracing measurements), over the course of the patient's involvement in the study. This information will help to determine if the prototype device can provide information about wound re-modeling in real-time that is indicative of wound healing (i.e., decrease in wound size and thus wound closure). Furthermore, the green fluorescence channel in the images obtained using the device will provide information about the wound margin (e.g. circumference).
Study wounds will be grouped according to whether or not bacteria were isolated and identified during microbiological laboratory procedures. Those containing any amount of bacteria (regardless of species) will be categorized as bacteria-positive wounds (e.g. contaminated). Those that were negative for bacteria will be categorized as bacteria-negative wounds. The bacteria-positive and bacterial-negative wounds will be examined for differences in clinical signs of infection using Fisher's exact tests. Differences between the bacteria-positive and bacterial-negative groups with respect to microbial load and diversity of species will be statistically examined using t-tests for independent groups. An alpha level of 0.05 (two-tailed) will be employed. Sensitivity and specificity for detecting bacteria in wounds will be calculated both per-wound and per-patient (a measure of ability to detect and correctly diagnose bacterial contamination in the wound). Furthermore, statistical analysis of multiple variables will be used to determine sensitivity, specificity and predictive value of the fluorescence imaging device for detecting (white light-occult) bacterial infection in wounds compared with standard methods.
Bacteriology results will be pooled for each wound type across the total patient population and scored based on CFU data in order to determine the percentage of wounds that were contaminated, colonized, critically colonized and infected. Statistical analysis will be used to determined degree of correlation between fluorescence intensity and the level of bacterial presence progressing from bacterial balance to bacterial damage in each chronic wound type.
This data will be used to determine whether fluorescence imaging can differentiate whether a given wound type is in bacterial balance (contamination with organisms on the surface or colonization with organisms in the tissue arranged in micro colonies without causing damage) or bacterial imbalance (critical colonization and infection), which is of primary importance to healing.

Potential Impact
We believe that this study will aid in developing an important new imaging tool for real- for chronic ulcers is expected to increase, thus the need for advanced wound care will increase.
Our study aims to develop PRODIGI™ as a new medical device to directly impact patient care by allowing cost-effective monitoring of chronic and acute wounds in real-time for