Case study: LED fluorescence microscopy
In September 2005, an expert meeting at WHO was organized to determine the policy recommendations for expanded use of fluorescence microscopy (FM), based on three systematic reviews commissioned by FIND and the WHO Special Programme on Research and Training for Tropical Diseases (TDR). These included over 50 publications in the peer reviewed literature and provided convincing data that, in experienced hands, FM was faster and an average of 10% more sensitive than conventional Ziehl-Neelsen (ZN) microscopy, with no loss in specificity. However, data on the use of FM microscopy outside of reference centres was found to be inadequate.
The relatively recent advent of low-cost ultrabright light emitting diodes (LEDs), made possible the development of inexpensive LED-illuminated microscopes that could be used in place of expensive mercury-vapor bulb solutions.
- A technical summary: Ely Silk, LED Fluorescence Microscopy in theory and practice, The Microscope, Vol. 50, 2/3, 101-118, McCrone Publishing, 2002
- The feasibility of using LEDs to adapt existing expensive fluorescent microscopes was proven by Anthony and colleagues: Light emiting diodes for auramine O fluorescence microscopic screening of Mycobacterium tuberculosis, Int J Tuberc Lung Dis 2006, Anthony et al.
Concept and development
FIND decided that the most useful approach would be the creation of a new microscope that could perform both fluorescent and brightfield applications. In accordance with FIND’s business model, this technology must be of the highest quality, but also affordable for high-burden countries.
Carl Zeiss MicroImaging LLC is a worldwide leader in optics and microscopy, with high-end LED-based fluorescence systems already in the pipeline and the capacity for global distribution. After signing an agreement in 2006, Zeiss and FIND began the co-development of Zeiss’ Primo Star iLED microscope, with some of the following product specifications:
The Primo Star iLED microscope
- high-grade optics
- reflected rather than transmitted blue light for fluorescent applications
- battery power on AC failure
- easy switching between fluorescent and brightfield viewing
- utility outside a darkroom
- >10,000 hour bulb life
- inexpensive bulb change
- LED and filter sets that would allow use of Auramine O and acridine orange for TB and parasitologic examinations
Early prototypes of the microscope were developed by the end of 2007 and exhibited at the IUATLD World Congress in Cape Town. A design-locked manufacture version was ready in the second quarter of 2008.
While FIND carried out its own product development, several other LED-based fluorescence microscopes for TB detection entered development or became commercially available. In March 2008, FIND held a meeting in Geneva on LED Fluorescence Microscopy, which was attended by a number of scientists, clinicians, and public health specialists working in this field.
The purpose of the event was to harmonize the strategies for evaluating the performance and impact of the different microscopes or microscope adaptors available. FIND shared its preliminary plans for demonstration projects in the hope that other investigators might help address some of the technical challenges, and that similar approaches might be used to evaluate the various tools. This would allow a harmonized set of data to be submitted to the Stop TB Department of WHO and its expert committees for review.
The feasibility of using the Primo Star iLED prototype for TB detection was determined in two laboratories experienced in conventional fluorescence microscopy. A set of study slides with 30-50% positivity rate was generated prospectively from patients undergoing diagnostic evaluation and stored until culture results were available. These slides were used in blinded studies assessing several factors, including:
- performance of the iLED microscope in comparison to the more expensive standard FM scope
- necessity of a darkroom
- suitability of the iLED excitation wavelength for alternative stains (Auramine or Rhodamine) and counterstains (KMnO4 or Methylene blue)
- reading time per slide
- speed of fading of stored slides
- appraisal of the usability and technical suitability of the microscopes by the laboratory technicians.
Once feasibility targets were met, the final Primo Star iLED manufacture series of the microscope was evaluated for performance in the detection of TB in sputum specimens in a laboratory-based, blinded, multi-center study. Panels of slides with specific grading of positivity were created by a single facility and shipped to 6 reference laboratories with prior experience in fluorescence microscopy that carried out the study. The endpoints of the study were:
- Sensitivity of iLED in smear positive panel slides compared to light microscopy and conventional fluorescence microscopy
- Specificity in smear negative slides compared to light microscopy and conventional fluorescence microscopy
- Assessment of technicians’ appraisal of FIND/Zeiss microscope in terms of ease of use, maintenance, design and comfort, robustness, contrast, brightness, etc.
- Assessment of the adequacy of 200x vs. 400x magnification for slide screening.
Once performance targets were met in the evaluation study, a demonstration project was initiated. This study was carried out in coordination with national TB control programmes (NTCPs) in high-burden TB countries that have a pre-existing agreement with FIND at national level (MOU with NTCP and/or MOH) and a local presence (either FIND or an implementing partner). Two countries were selected that had a high prevalence of HIV.
36 primary sites and 20 extension sites were chosen for this large demonstration study. Site selection was based on the rate of smear-positivity, the training of microscopists, the volume of work, the reliability of AC power (sites intentionally selected with intermittent power supply), interest in the project, and accessibility of study sites to supervisory visits. At least 20 centers were included without any prior experience of fluorescent microscopy to determine the performance of LED FM in these settings.
For all sites, baseline performance data was collected using the standard ZN procedures over a month long pre-study period. Following a 5 day training session on the Primo Star iLED microscope, sites initiated a validation phase. During this phase, all slides are screened with the iLED microscope and then confirmed on a daily basis using a conventional fluorescent microscope by a supervisory site. Patient care is based on the results of conventional FM examination.
Proficiency testing panels are also used to assess the study sites. Sites can only participate in the demonstration phase if they show acceptable performance at the validation phase. This is defined by a 95% result concordance between the demonstration site and the supervisory site, an acceptable quality of Auramine stain in 100% of panel slides examined and a non-significant number of false results in the proficiency testing panel.
Sites that pass the proficiency testing continue on to a 3 month phase in which the iLED microscope is used as the primary screening tool. Stored slides are rechecked throughout this phase, and results are compared to those from the pre-study period (using ZN microscopy). In some sites, additional operational questions are being answered. The primary and additional study objectives are listed below.
Primary Objectives (Basic Protocol)
- To assess the feasibility of implementing Primo Star iLED for TB diagnosis at microscopy centers without prior experience with fluorescence microscopy in low- to moderate-income settings and to identify barriers to implementation
- To determine the false positivity and negativity rate of LED fluorescence reading compared to a ZN baseline and to results from the supervisory site
- To determine the development of false positivity and negativity rates of LED fluorescence reading over time (with increasing experience)
- To assess the impact of this implementation on daily workload and case detection rates for low, middle and high-volume settings
- Determine lab technicians’ appraisal of Primo Star iLED
- To evaluate detailed costs associated with LED-based fluorescence microscopy in comparison with conventional methods
Secondary Objectives (Extended protocol - FIND partners)
- In collaboration with other groups and in preparation for WHO STAG submission, establish comparative performance data for alternative LED-based approaches
- To identify minimal training needs and develop training modules accordingly
- To identify the optimal fluorescence staining method and ensure continuous and affordable supply
- To assess effects of fading speed on external quality assurance by rechecking
- To assess whether combining LED fluorescence microscopy with concentration methods such as bleach sedimentation further increases sensitivity
The size and complexity of the Demonstration study reflects the importance of having true performance data from sites with no prior experience in fluorescence microscopy. In the coming years, LED-based FM is likely to see a marked increase in use and it will be critical to understand the training and supervisory needs for peripheral microscopy sites using this technology, as compared with conventional brightfield examination of ZN-stained slides.