Final Report Summary - AUTOCAST (Automatic Cancer Screening Based on Real-time PCR)
The AUTOCAST project has developed a new panel of messenger Ribonucleic acid (mRNA) biomarkers for early detection of cervical cancer, a novel low cost automated real-time Polymerase chain reaction (PCR) / probe technology, in a microarray biochip format with a corresponding automated detection instrumentation for use as a rapid 'point of care' diagnostic device in both a clinical and laboratory setting. Cervical cancer and its associated virus, Human papilloma virus (HPV) is the model system which was used in the current project to exploit the potential of the novel real time PCR technology, namely the Mediator probe assay (MPA) to detect panels of multiple biomarkers, both viral and host cell markers, within a single test.
Within the project, AUTOCAST has developed the following:
1. A cervical mRNA expression biomarker set for sensitive and specific detection of CIN2+ lesions in cervical smear specimens which has been validated in a large cohort of colposcopy referred patients with histologically confirmed disease status. The complete clinical validation of the selected biomarkers panel was performed on standard real time PCR instrumentation and yielded excellent results. For the detection of CIN2+ cases the biomarker panel specificity (0.93) and sensitivity (0.89) were better than standard screening methods including high risk HPV Deoxyribonucleic acid (DNA) tests (specificity 0.85 sensitivity 0.88) and / or cytology alone with ASCUS+ cutoff (specificity 0.83 sensitivity 0.70) or with LSIL+ cutoff (specificity 0.97 sensitivity 0.54) respectively.
2. A novel real time PCR detection technology for use in solid phase microarray formats called the MPA has been developed and its proof of principle was demonstrated for both liquid and solid phase chemistries. MPA developed within project combines the enormous multiplex capability of microarrays together with the sensitivity of real-time PCR. Additionally the assay enables the use of a universal microarray whilst any gene could be detected by changing the primers and probes of the PCR.
3. A unique PCR cartridge was designed which includes an optical and fluidic chip, with 8 channels, each channel suitable to multiplex up to 20 gene targets. The material for each chip was selected fulfilling production, thermal, optical, chemical, biomolecule immobilisation and PCR biocompatibility requirements. Chips have been produced by injection moulding and several bonding and coating options have been evaluated.
A low-cost total internal reflection fluorescence and epifluorescent PCR cycler / scanner has been produced which has automatic data handling capabilities. Partial integration of these various technologies was tested on clinical specimens using the liquid phase, MPA and PCR scanner technology and targeting the application of the developed biomarker panel.
Project context and objectives:
Globally cervical cancer incidence rates have increased from 378 000 (256 000 - 489 000) cases per year in 1980 to 454 000 (318 000 - 620 000) cases per year in 2010, a 0.6 % annual rate of increase. Cervical cancer death rates have been decreasing but the disease still killed 200 000 (139 000 - 276 000) women in 2010, of whom 46 000 (33 000 - 64000) were aged 15-49 years in developing countries. It is usually preceded by a long phase of pre-invasive disease called Cervical intraepithelial neoplasia (CIN). This pre-invasive phase is characterised microscopically as a range of events progressing from cellular atypia to various grades of dysplasia, including CIN before progression to invasive carcinoma. This precursor phase is generally asymptomatic, and can occur over a long period of 10-20 years.
CIN is the single most important aetiological agent in the pathogenesis of cervical cancer and pre-cancer. About 15-20 types are associated with cervical cancer, of which CIN16, 18, 31, and 45 accounts for 80 % of cervical cancers. CIN vaccination has now become part of cervical cancer screening and prevention strategies worldwide. There are two prophylactic vaccines currently available which protect against two specific high risk CIN types 16 and 18. With the introduction of CIN vaccination, the landscape of cervical pre-cancer will change over time. While the incidence of abnormal smears and high grade disease will decrease over time, it is anticipated that in the future, the incidence of some type of low grade abnormalities may increase as a consequence. Furthermore, in vaccinated populations, the lower prevalence of disease will directly impact on the performance characteristics (i.e. positive predictive and negative predictive values) of current diagnostic tests.
There are no specific clinical features or symptoms which indicate the presence of CIN. Initial diagnosis is usually made by cytological analysis of a Pap smear specimen. Cytological analysis of PAP smears is the current method employed by cervical screening programmes for the initial diagnosis of abnormality. The result of the PAP smear is what refers the women for additional screening in colposcopy. However it is well recognised that there are limitations to the existing PAP smear screening procedures. These include a significant incidence of false positive and false negative results. Efforts to improve the specificity and sensitivity focus on triage of cytology with CIN testing, which certainly can improve the sensitivity but issues with specificity still remain. As eluded to above in CIN vaccination populations, the rate of abnormal cytology will decrease, this will mean it will be more challenging for cytotechnologists to review cervical cytology samples. It is likely therefore that CIN testing will become the primary screening event in vaccinated populations. There remains a lot of debate over the best method for triaging women who test CIN positive into high and low risk groups. In this context biomarkers will be an important consideration. Currently histology is the gold standard procedure in terms of disease diagnosis. However histological diagnosis of CINs is complicated by a variety of cellular changes associated with inflammation, pregnancy and / or atrophy. These changes may mimic precancerous cervical lesions, thereby making traditional cervical histology approaches, subjective and prone to variability. This is reflected in poor inter-observer agreement between pathologists. In particular, the differential diagnosis between immature squamous metaplasia and CIN1 / 2, or between low-grade (CIN1) and high-grade (CIN2 / 3) lesions, tend to be difficult. The limitations associated with histological diagnosis, impact greatly on clinical trials in the area of cervical cancer pre-cancer biomarker validation, CIN vaccination, and novel therapies, as histology is usually the end point in terms of disease. Accurate grading of CIN lesions is paramount for clinical management of patients because CIN 1 and CIN 2-3 lesions are treated differently. Furthermore, as histological diagnosis of cervical biopsy ultimately determines the decision to treat, combined with the low positive predictive value of the smear test which initially refers women into colposcopy, often results in unnecessary treatments. The estimate of this burden has been assessed in several studies, where up to 20 % of low grade lesions were upgraded and 26 % of high-grade lesions were downgraded after histological review. The issues relating to accurate diagnosis of CIN can result in over- or under-treatment of the lesion. Moreover, in early CIN lesions, the progression rates to high-grade lesions are low (CIN1 and 2 to CIN3 is 9-22 % and to invasive cancer is 1-5 %). This effectively means that many women unlikely to progress are being over-treated and followed up unnecessarily, causing a huge burden on healthcare systems. Collectively, this emphasises the need for specific biomarkers to aid objective CIN lesion grading, to identify true high-grade cervical disease, and to increase the specificity and PPV of disease detection.
Molecular methods (especially nucleic acids based assays) are powerful diagnostic tools for rapid detection of specific gene sequences in clinical samples. However, molecular diagnostic tests in general, such as the multiplex polymerase chain reaction (PCR), DNA sequencing and microarrays, are still labour intensive, time consuming and complex investigations. To overcome these limitations and to enable simpler use of these methods for medical diagnostics, the whole process of molecular tests requires simplification and revision with miniaturisation and automation presenting a great potential for the integration of molecular assays in an easy-to-use and flexible point of care diagnostic system. In this project, we have developed a novel low cost automated real time PCR / probe technology, in a microarray biochip format with corresponding automated detection instrumentation for use as a rapid point of care diagnostic device in both a clinical and laboratory setting.
Cervical cancer and its associated virus, CIN is the model system which has been used to develop this novel automated cancer screening technology. The uniqueness of the current project is the potential of the novel real time PCR technology to detect panels of multiple biomarkers, both viral and host cell markers, within a single test. This will impact enormously on current cervical screening programs particularly in the context of CIN vaccination. Incorporation of these biomarker panels into cervical screening programs has the potential to stratify patients based on their CIN subtype, defining different treatment cohorts, resulting in improved patient management and reduced costs to healthcare providers.
The technology described in the presented approach is based on a microfluidic disposable cartridge enabling real-time PCR amplification and detection. The disposable cartridge is processed by a small integrated low cost thermo-cycler and reader system which can perform the diagnostic test automatically. The diagnostic test relies on a panel of biomarkers, which are detected within the cartridge to allow for multi parameter analysis. The detection is based on the most sensitive Total internal reflection fluorescence (TIRF) method and has been modified to facilitate the the more common epifluorescent technique. By this, the sensitivity and specificity of detection can be improved and there is a potential to generate quantitative results.
The novel real time PCR approach, which incorporates a novel type of MPA, is applicable to a large variety of diagnostic markers and PCR-based tests. This approach enables the use of a single generic microarray for multiple purposes. Thus the high volume fabrication of the disposable cartridge hosting the microarray can be largely facilitated and commercial uptake can be fostered. The developed technology has the potential to eliminate the bottlenecks of current immobilised real-time chemistries such as immobilised molecular beacons. The developed system and assay technology can not only be used in the medical field for cancer research and diagnostics, but will also be applicable for food safety, environmental monitoring or homeland security if a suitable biomarker panel is adopted. The applicability and performance of the developed prototype system in the AUTOCAST project has been demonstrated for detection of HPV (CIN) and mRNA biomarkers in cervical cancer.
With the recent introduction of the Gardasil CIN vaccine (Sanoffi-Pasteur) in Europe, the combined use of cervical screening and a CIN prophylactic vaccination programme heralds a new era in cervical cancer prevention. The diagnostic device developed within the current project will advance and complement international cervical cancer vaccination and screening programs, through its ability to stratify patients based on their CIN and biomarkers status, in a point of care setting. As part of the project a clinical study has been performed on clinical specimens to validate the full system i.e. the new MPA for CIN detection in combination with the microfluidic cartridge technology. The novel system has been benchmarked against the gold standards currently used in the clinical laboratories.
Any new device to be used in primary screening protocols should include an aim to increase sensitivity and specificity and reduce inappropriate high referral rates from a targeted screening population. The currently available tools used in cervical screening (PAP smears, hybrid capture CIN testing) in general have low positive predictive value and variable sensitivity (5). This has hindered widespread extension of CIN screening programmes.
Nevertheless, the combination of cytology and CIN testing significantly improve the accuracy of primary cervical screening. Therefore, there is a need for a sensitive and specific, high-throughput diagnostic device which will facilitate CIN and biomarker population screening. The attractiveness of a microchip device with multiple reporter channels facilitating sub-type CIN analysis and other viral (load) and human biomarkers in a unified platform represents a major step forward in cervical cancer screening and diagnostics. Such a platform using a microarray device has the potential for development in other diagnostic systems (including other cancers), biosafety platforms and food technology. Microarrays are in principle ideally suited as miniaturised assay platforms for such a multi parameter diagnostic device. What is required is the coupling of a multiplex real-time amplification step with detection on an array platform, heretofore not possible. This proposal attempts to develop a real-time PCR based array technology utilising novel mediator probe chemistry. The microchip will use CIN sub-type analysis and other human biomarkers as a proof of concept for this technology approach.
Main objectives
The aim of the AUTOCAST project was to develop and to validate a point of care (POC) diagnostic system for cancer screening, combining the methods of real time PCR and DNA microarrays. The complete system will comprises:
- a combined thermal cycler and optical detection system to perform PCR cycling and microarray detection simultaneously;
- a microfluidic cartridge containing light guiding components to enable Attenuated total reflection (ATR) and the immobilised microarray within a reaction chamber, which is amenable for PCR cycling and optical readout by TIRF ;
- a novel type of assay protocol (the MPA) to enable screening of any nucleic acid by a single universal/ generic microarray ;
- a new combinational approach for cervical cancer screening including both CIN and human gene biomarker testing;
- a clinical feasibility and validation study to benchmark the performance of the new test to existing diagnostic platforms with respect to CIN diagnostics.
Project results:
According to annex I objectives the following were achieved:
- A combined thermal cycler and optical detection system to perform PCR cycling and microarray detection simultaneously
An automated PCR microarray analyser has been developed within WP4 of the project AUTOCAST. The system includes an optical readout based on TIRF and epifluorescence and a temperature control unit for PCR cycling. The whole system is controlled by electronics and software for a fully automated operation. The software has several updated version according to the project requirements and also to include several new features. Instruments were used at GenoID, Budapest and Trinity College Dublin, respectively. The third one remained in Freiburg to be used by IMTEK and Fraunhofer IPM. D4.4 deliverables and the associated tasks are completed. The AUTOCAST real-time PCR microarray analyser BCR 603 was tested and characterised. The systems redesigned later (prototype) for the clinical validation was also characterised. The analyser's specifications have been met according to the specification of deliverable 4.1 and it is used the characterisation methods in deliverables D4.2. D6.1 D6.2 and D6.3 and the associated tasks are completed. The D6.4 was joined with D6.2 because the delayed chip production, the reagent and performance validation of the demonstrator design was no longer feasible, so only the prototype system was validated.
- A microfluidic cartridge containing light guiding components to enable ATR and the immobilised microarray within a reaction chamber, which is amenable for PCR cycling and optical readout by TIRF
For cartridge demonstrator the adequate plastic material has been determined that will withstand the operation conditions. The AUTOCAST Cartridge consists of the optical and the fluidic chip. The optical and fluidic chip had been delayed due to injection moulding problems, but later successfully completed. For cartridge prototype several bonding techniques were used, laser welding and adhesive bonding procedures were evaluated. The compatibility with mass production of the laser welding process makes it a very promising candidate as a final bonding technique; however it was incompatible with oxygen plasma treatment, which in turn was necessary for functional probe immobilisation of real-time PCR probes. Several contingency measures were introduced with limited success (selective coating, different non oxygen plasma dependent immobilisation methods, thermal bonding). Adhesive bonding was partly incompatible with the PCR process, and also it cannot support the production of larger numbers of chips, which are necessary for the clinical evaluation. Thermal bonding was also incompatible with oxygen plasma treatment. With delay, however, the D5.2 D5.3 D5.4 D5.5 and D5.6 deliverables and the associated tasks are completed.
- A novel type of assay protocol (the MPA) to enable screening of any nucleic acid by a single universal/ generic microarray
The whole assay integration of the singleplex MPA is presented. Five different concepts were evaluated to detect a liquid phase target gene spatially resolved on a solid support. All concepts are based on the Mediator Probe cleavage accompanied by the release of the Mediator. The different approaches comprise different enzyme classes or significantly modified oligonucleotides like multiple fluorophores or inversed internal linkage. The presented detection techniques were developed and evaluated in regard of manufacturing aspects. Regarding feasibility the concept consists of an immobilised hair-pin shaped oligonucleotide with fluorophore and quencher moieties ('Universal Reporter'). A prerequisite for intensive testing and optimisation is the immobilisation of the oligonucleotides using a thermostable coating of the chip surface (cooperation partner Austrian Institute of Technology), the sealing of the microfluidic and optical part of the cartridge by laser welding (cooperation partner Jenoptik Polymer Systems) was not fully achieved during the timeframe of the project. The low reproducibility of the results hampered our efforts to fully deliver D3.3. D3.1 is done partly and discontinued (this was a risk aversion contingency route), D3.2 deliverable and the associated tasks are completed.
- A new combinational approach for cervical cancer screening including both CIN and human gene biomarker testing
A new expression biomarker set for cervical cancer screening has been established. The established biomarker geneset fulfils the criteria to complete the WP D2.3 but we had postponed the due date of the D2.3 to investigate some other alternative possibilities, and establish contingency genesets for further use. The D2.3 delivered significantly improved results. For the detection of CIN2+ cases the biomarker panel specificity (0.93 95 % CI 0.90-0.96) and sensitivity [0.89 95%CI 0.86-0.91] were better than the values of HR-CIN test (specificity 0.85 95 % CI 0.81-0.88 sensitivity 0.88 95 % CI 0.85-0.90) or cytology alone with ASCUS+ cutoff (specificity 0.83 95 % CI 0.77-0.88 sensitivity 0.70 95 % CI 0.66-0.73) or with LSIL+ cutoff (specificity 0.97 95 % CI 0.93-0.99 sensitivity 0.54 95 % CI 0.51-0.55). Bayesian point estimates of positive (PPV) and negative (NPV) predictive values for a hypothetical, natural history based, screening population with 1 % of CIN2+ prevalence are([PPV: 0.12 0.05 0.04 0.17; NPV: 0.999 0.998 0.996 0.995) respectively. Favouring high screening specificities to keep screening costs low for special settings an optimal cut-off value for the biomarker panel test was selected. This was achieved at a compromise of sensitivity for CIN2+ (0.50 95 % CI 0.48-0.50) however the point estimate of PPV improved significantly to 0.55 and the corresponding NPV is still high (0.994). Nevertheless this cut-off still provides high sensitivity for the detection of cervical carcinoma (0.83 95 % CI 0.70-0.86). D2.1 D2.2 D2.3 deliverables and the associated tasks are completed.
- A clinical feasibility and validation study to benchmark the performance of the new test to existing diagnostic platforms with respect to CIN diagnostics
The overall output of this deliverable is to provide a biobank which consists of a comprehensive collection of well characterised cervical cytology smear samples collected from a defined population of women with cervical abnormalities. The AUTOCAST real-time PCR microarray analyser, novel MPA chemistry and 3 prototype chip formats have been successfully demonstrated for detection of CIN16 and mRNA biomarkers in cervical smear specimens. Also we have successfully demonstrated a multiplex CIN PCR detection, with simultaneous amplification of 14 high-risk CIN genotypes in a single channel. To access its value a case control study using Hungarian and Irish cervical screening populations was set up. Cases comprised 750 women who developed CIN2+ or worse confirmed by histology, and were stratified by age group (under 29 and over) and disease grade (CIN2, CIN3, and carcinoma). Controls (n equal to 250) were women who identified disease free by cytology and were the same age on average as cases, the controls were also stratified by age. Biomarker reverse transcriptase PCR measurements using routine exfoliated cytological samples, high-risk CIN testing, cytology and/or histology were performed and used to evaluate different protocols of testing including combinations of the tests with different test cut-offs. The Bayesian estimates of critical parameters of test performance were calculated and compared for the different stratified groups. D7.1 D7.2 and D7.3 deliverable and the associated tasks has been delivered and reported on schedule.
Detailed scientific and technological results
Here we present a short summary of the results we have generated for each of the specific tasks across the 7 WPs. Overall we present the development of a new diagnostic cervical cancer in vitro diagnostic device capable of CIN and mRNA biomarker detection in clinical samples.
WP1 - Selection of biomarker panel and device requirements
Preanalytical constrains
Within this task we defined the optimal sampling medium for processed and use with the AUTOCAST system. The sampling type and medium we have chosen is liquid-based cytology smears specimen collected preferably into ThinPrep PreservCyt, solution. This will complement another project (SPE_SAFE) we have running alongside AUTOCAST, funded through a Hungarian national grant which is developing a combined sampling and extraction device for RNA and DNA within the same timeframe as the AUTOCAST project. This device is designed to provide long ambient sample stability. Common exploitation is proposed for both foregrounds.
We have also defined the required sensitivities we require both for signal intensities of fluorescent detection and PCR analytical sensitivities, in both cases the projection is that the prototype AUTOCAST device should be capable of achieving similar sensitivities to conventional liquid phase real-time PCR reactions. The reaction volume which we will use in the system is 5 ul per detection channel, comprising of 4 ul reaction mastermix and 1 ul nucleic acid sample volume. In order to achieve good analytical sensitivities for the whole sample volume the SPE_SAFE is expected to deliver a highly concentrated nucleic acid solution. The current AUTOCAST chip design has 8 channels with 20 immobilisation areas in each channel for the detection probes. The current instrument is capable of reading at 2 wavelengths so the maximum number of the reactions is 20 reaction in one channel, altogether 160 reactions per chip. While the complete multiplex real-time PCR reaction was not demonstrated within the timeframe of the project, the initial proof of concept studies have demonstrated the feasibility of this high level multiplexity approach. . The results have demonstrated mediator probe surface detection, multiplex CIN amplification and sensitive detection of liquid phase biomarker expression. Nevertheless for the proposed CIN multiplex detection (for 20 genotypes), typing and viral load determination (quantitative detection), the established liquid phase mediator probe reaction is capable of working at a multiplicity of 2. The 8 channels are divided between positive control, negative control and duplicated target detection for both biomarker and CIN detection tasks. The plan was for the other wavelength to be used for spotting control. A universal thermal cycling profile has been developed for the MPA. The overall aim is that the cycling time is kept below one hour to deliver a viable point-of-care instrument, this is achievable with this universal setup, however the current multiplex CIN protocol takes a bit longer. The detection dynamic range is large enough to cover several magnitudes and does not limit the overall detection performance. The software is already available for control of detection and thermal cycling, and will deliver background and negative control corrected evaluated amplification results.
Evaluation of required technologies
For benchmarking, development of reference materials and reactions was carried out: Genoid supplied BF and IMTEK with CIN reference materials, for the biomarker reactions Genoid established a TaqMan based reaction format for two-site evaluation of the biomarker genes (TCD and Genoid). IMTEK made reference reactions and material available for Genoid and TCD regarding the mediator probe chemistry including probe designs for both mediator and molecular beacon reactions. Regarding material selection and technological decisions, evaluation of the possible immobilisation chemistries in accordance with the optical, thermal, chemical, biochemical features of the cartridge material was carried out. Physical design requirements were set as described above. To ensure a low cost device, the technologies are carefully designed, low cost, high productivity injection moulding, laser welding, two-step (chemistry and probes are spotted in two consecutive step in the same technology step), top-spot immobilisation was the planned approach.
Selection of biomarker panel
Selection of genes for the biomarker panel was carried out. The established classification biomarker panel of 20 genes were initially validated on an independent set of samples from TCD biobank (136 samples) and at TCD using independent instrument and personnel and later was validated at Genoid on larger population. All of the cases comprised 750 women who developed CIN2+ or worse confirmed by histology, and were stratified by age group (under 29 and over) and disease grade (CIN2, CIN3, and carcinoma). Controls (n equal to 250) were women who identified disease free by cytology and were the same age on average as cases, the controls were also stratified by age. For the detection of CIN2+ cases the biomarker panel specificity (0.93 95 % CI 0.90-0.96) and sensitivity (0.89 95 % CI 0.86-0.91) were better than the values of HR-CIN test (specificity 0.85 95 % CI 0.81-0.88 sensitivity 0.88 95 % CI 0.85-0.90) or cytology alone with ASCUS+ cutoff (specificity 0.83 95 % CI 0.77-0.88 sensitivity 0.70 95 % CI 0.66-0.73) or with LSIL+ cutoff (specificity 0.97 95 % CI 0.93-0.99 sensitivity 0.54 95 % CI 0.51-0.55).
WP3 - Assay chemistry
A novel assay chemistry which allows simultaneous real-time amplification and detection of multiple targets in parallel was developed. This new technology is called MPA. This mediator probe mediated detection enables the fabrication of universal oligonucleotide arrays for simultaneous real-time PCR detection. Several approaches using molecular beacon chemistry in solid and liquid phase were also evaluated as a contingency to reduce the risk, should the MPA concept fail. These were later abandoned once the MPA proof of concept was demonstrated. Small volume real-time liquid phase reaction
Low volume real-time liquid phase reaction were tested for molecular beacon chemistries (using red wavelength excitation/emission), which was comparable to normal volume reaction. The optimal polymerase and buffer pair was selected to achieve effective mediator probe cleavage. Liquid-phase mediator probe cleavage was demonstrated.
Solid phase adaptation of molecular beacon assay
Solid phase molecular beacon chemistry was a contingency measure and was published to be successfully implemented in other PCR reactions, so this was discontinued as a task.
Development of mediator probe chemistry
Successful implementation of mediator probe technology was achieved after investigation of several assay principles. The mediator hybridises at an immobilised hair-pin oligonucleotide exhibiting a specific mediator complementary site. The polymerase extends the mediator whereby the 5-terminus of hair-pin oligonucleotide is digested and the quencher molecule is cleaved off. The polymerase elongates the primer until the polymerase activity is inhibited by an abasic site (optional) or by steric hindrance due to the hair-pin or internal linker. The polymerase dissociates from the oligonucleotide, the elongated Mediator is available for the next hybridisation event. Assay performance is correlated to immobilisation efficiency and reproducibility of the 'Universal Reporter' molecule. Several concentrations of the Universal Reporter were tested. The concentration of the Mediator Probe was optimised to 200 nM. Increasing the concentrations neither improved the sensitivity nor increases the fluorescence signal in the assay. The data show a clear correlation between the fluorescence signal plot and the initial DNA copy number. Target DNA down to 10 copies per reaction can be detected. In approx. 95 % of the controls no signal increase was observed showing the specificity of the demonstrated MPA. The cleavage of the Universal Reporter is indispensable for a successful accomplishment of the MPA, especially on solid phase. This implies several requirements for the substrate material, coating and immobilisation chemistry:
- thermostability;
- selective orientation and immobilisation via the linking group;
- non-interfering fluorescence interaction;
- overall feasibility for the assay implementation.
Proof-of-concept studies for the MPA utilising immobilised 'Universal Reporter' molecules within a functionalised microwell ('global coating') were reported in Delieverable D3.2. To transfer the MPA technology to a polymer cartridge with spatially resolved features a simplified setup is presented. An array comprising a limited set of different oligonucleotides is printed on a polymer slide. Furthermore, the polymer slide is bonded with a structured substrate to a microfluidic cartridge. After bonding (Biofluidix) the cartridge was washed (according to a protocol provided by AIT). After heat-activation of the polymerase and cooling down to the annealing temperature, the Mediator hybridises to its complementary sequence site within the Universal Reporter sequence and is processed by the polymerase. The 5' terminus of the 'Universal Reporter' is degraded resulting in an increased fluorescence level. Due to a well-considered array layout, influences like coating, printing buffer, and applied oligonucleotide concentration on thermostability, selective orientation, quenching effects and assay feasibility can be identified and quantified. The intensities of the immobilised oligonucleotides differ significantly. As the control oligonucleotide is modified with a fluorophore the highest signal intensity is detected. The duplexing capability was demonstrated. Two different Mediator molecules trigger signal generation at a specific 'Universal Reporter'. These findings allow the conclusion that principally higher multiplexing degrees are feasible. However, experiments combining target amplification and solid phase detection were not achieved within the timeframe of the project.
WP4 - Real-time PCR microarray analyser
The TIRF based real-time thermocycler array reader (real-time PCR microarray analyser) has been developed and tested. The instrument is capable of reading microarrayed fluorescent signals in real time. In parallel to the assay development, the instrument has been optimised for the diagnostic microarray readout including on chip PCR and real time detection. The partners required instruments for their work at an early stage within the project. Therefore, two stages of development were planned. In a later project stage, the final version of the instrument (second generation) was built up, taking into consideration the experience with the first generation. The second generation fulfils the criteria for clinical validation.
Definition of requirements
The system specifications were defined as a first step of instruments development. Aspects of fluidics and production techniques were taken into account. The requirements include the data of optical detection such as number of dots (20 per channel and 8 channels), area to be read out (8x10 mm), required Limit of detection (LOD) (comparable to state-of-art scanners), wavelength ranges. The temperature requirements for PCR cycling have been defined to make possible amplifications below one hour time limit to support point-of-care applications. The cartridge handling includes a locking mechanism using drawer design and the user interface is a computer program, which makes possible to manipulate all important experimental parameters.
Definition of test procedures for optical and thermal assessment
The definition of test procedures has been defined regarding spot sises surface, the dynamic range and the optical cross talk. Thermal properties include the accuracy and stability of temperature control. All these parameters were measured during analytical evaluation of the prototypes devices.
Demonstrator instrument
On the basis of the specifications, instruments were realised. Simple and automated way to couple the light in after inserting the cartridge into the instrument has been successfully implemented, special optical coupling structures have been designed, tested and realised in the polymer technology. The coupling efficiency and its variations have been characterised and optimised. Homogeneous illumination of the readout area has been achieved, which is necessary for a comparable detection of fluorescence intensity. A low cost, the sensitive detection has also been realised by very simple means in order to keep fabrication costs of the system extremely low. An optimised high aperture imaging optics including spectral filtering has been developed on the basis of commercially available components. High-speed temperature cycling has also been implemented. The instruments have been developed, CE marked and analytically validated. Due to the delay in the delivery of the final AUTOCAST chip, complete overall validation of the system has not been completed to date.
Prototype instrument
In a later project stage, the final version of the instrument (second generation) was built up, taking into consideration the experience with the first generation. In parallel to the instruments development, the cartridge was developed. Therefore, the first instrument operates with standard glass slides, the second generation works with the final cartridge. The second generation was used for clinical validation and proof of concept studies These instruments are operated by clinical lab personnel, requiring a high degree of user friendliness and robustness. Training and maintenance task are necessary to guarantee a successful validation process. Instruments were used at GenoID, Budapest and Trinity College Dublin, respectively. The third one remained in Freiburg to be used by IMTEK and Fraunhofer IPM. The AUTOCAST real-time PCR microarray analyser BCR 603 was developed, tested and modified. The systems redesigned for the clinical validation was characterised. The analysers specifications have been met according to the specification of D4.1 and it is used the characterisation methods in D4.2. D6.1 D6.2 and D6.3 and the associated tasks are completed. The D6.4 was joined with D6.2 because the delayed chip production, the reagent and performance validation of the demonstrator design was no longer feasible, so only the prototype system was validated.
WP5 - Microfluidic cartridge design
Design and fabrication of the disposable microfluidic cartridge including the microarray for multi parameter analysis has been achieved. Establishing a fabrication technology and supply of cartridges to the other work packages throughout the project were delivered and optimisation of the performance in terms of thermal, fluidic and optical properties were carried out.
Selection and validation of materials
Adequate plastic material has been determined which withstands the later operation conditions. The material has been qualified in functional tests regarding mechanical and optical characteristics (low autofluorescence and birefringence), resistance to chemicals and temperature resistance. Furthermore, the material has to meet the relevant criteria for an economic as well as a steady injection moulding process, e.g. in relation to the shape design (micro fluidic cartridge, micro array immobilisation) and sealing. The material has also tested regarding PCR and immobilisation conditions.
Cartridge design
The cartridge consists of an optical chip and a fluidic chip. Both of them are injection moulded. The optical chip incorporates an optical quality readout area and waveguide structures. The fluidic chip has eight detection channels and to sacrificial channels inlets and outlets. The sacrificial channels are applied to prevent clogging of detection channels due to gluing the components, which method is used before the laser welding is available. Before the PCR, channels must be sealed or closed, this has been established.
Construction and tooling and development of the moulding process
JOPS constructed tools for moulding both the optical and fluidic chips. Reproducible moulding of optical chip has been achieved. Production process optimisation contained the following key aspects:
- improvement of the tool operation;
- adaption of machine and systems engineering for realising the process requirements;
- defining a stable working point and process window for injection moulding and injection compression process.
The following research and work has been performed:
- integration of suitable heating and cooling systems with a minimum installation size and high efficiency for implementing the variotherm processes in the tools;
- taking account of the optical component requirements by maintaining the stiffness of the optical contour inserts;
- evaluation and installation of necessary draft angles in the area of the fluidic channels and coating of the tool inserts for reducing the component grip and guaranteeing a damage free ejection;
- researches for improving the polishing as well as the surface quality of the optical coupling and reflective surfaces;
- evaluation and adjusting the pre-stressing of the frame plates for ensuring the filling and compression operation;
- testing and optimising the overall function of the tools;
- purchasing, installation and testing applicable compression software for machines and hardware adaption;
- selection, purchasing, installation and testing applicable heating and cooling technology for realising the required variotherm process;
- evaluation the influence of the process parameters on the component quality.
The following parameters and machine settings had an impact on the process result: injection rate, injection and holding pressure, change-over-point to holding pressure, melt temperature, mould temperature or mould temperature profile, cooling time, clamping force and clamping force profile, stamping stroke, stamping start and stamping speed, development and qualifying process according QS 9000 and ISO 13485.
Optimised and qualified moulding process
To assure that a forming tool or mould together with the appropriate machine and the production process has the ability to manufacture tool and mould-bounded parts that meets the requirements of the AUTOCAST system, several steps were involved. Basic components of the qualifying process: assignment and release of manufacturing drawings, realisation of the tool or mould, risk assessment, qualification (installation qualification, operation qualification), initial sample testing, short-term process analysis (Ppk-Analysis), and production handover were performed. A statement regarding the achieved quality and stability of the process is possible with a Short-term process analysis (Ppk-Analysis). The chosen processes were evaluated and analysed accordingly. The single measurement values of the mechanical dimensions as well as of the birefringence resulted in very good to satisfactory PpK-values. Based on this, the capability of the chosen and analysed processes was demonstrated.
Chip pre-treatment and immobilisation chemistry
Providing a suitable coating is highly challenging as polymer surfaces behave differently compared to glass surfaces. Further, coatings which feature high immobilisation efficiency must not necessarily be suitable for the solid-phase MPA, especially when looking at specificity or thermal stability. Thus, coatings which require more complex (i.e. multi-step) application procedures which can complicate chip fabrication have been investigated. Finally, this led to the identification of potential coating candidates (termed advanced coatings).
Spotting and validation of the microarray
The printing of the assay-specific reagents on the immobilisation coating must fulfil two requirements. First, the printing concept itself must comply with the boundary conditions of the AUTOCAST cartridge design and allow for rapid chip production. Second, independent of the applied coating, the reagent droplets must be printed accurately into the respective channels. Both requirements are met as the current printing concept and instrumentation (TopSpot D) allows for the printing of up to 250 chips per day while ensuring that the reagent spots stay well inside the channels even if assuming worst case scenarios. The array quality (i.e. fluorescence intensity and S / N-ration) has been determined by conventional microarray hybridisation experiments.
Development of the sealing process
During the project, chips were mostly joined by adhesive bonding (please refer to D5.2) and by this process, strongly bonded chips could be produced with high yields. However, this process did only offer limited throughput thus limiting the number of chips that could be produced for the clinical trials. It was thus decided to develop a joining process based on laser welding which offers excellent scalability due to the short required process time per chip. Finally, a new method was developed which featured the best yields and lead to spotted and sufficiently welded cartridges. In this approach, the optical chip was first covered by a laser cut PSA (pressure sensitive adhesive) foil. Then, the complete coating process chain (i.e. plasma treatment and multiple wet chemical processing steps, please refer to deliverable D5.5) could be applied to these chips. It was demonstrated that this method works in principle and valid joining processes can be established this way, with occasional errors and interference effects to be eliminated.
WP6 - System integration
During the prototyping of the device (CIN biochip) it was essential to monitor the fulfilment of the requirements of the prototype and the final device. Analytical validation of the prototypes included: thermal, optical and fluidic validation, software validation and performance validation / cross validation.
Definition of test procedures and thermal and optical characterisation
Measurement of temperature with sensors was carried out. The specified temperatures should be reached in the reagents inside the channels of the chip. As during the PCR cycling no temperature measurement inside the channel is possible, the sensor for the temperature control is fixed on the surface of the peltier element. With PT100 (and other sensor types) the temperature inside the chip can be reached as specified. However the measured temperatures inside the chip are not reproducible in the specified range. Further improvement of the heat transfer between the instrument and the chip should improve the reproducibility. The ramp rates inside the chip have a peak rate of maximum 4 degrees of Celsius / second and an average of maximum 2 degrees of Celsius / second. To enhance the ramp rates several improvements were tested: the thickness of the chips is reduced, the temperature transfer was optimised and an enquiry for peltier elements with higher power was carried out. Despite the inaccuracy of temperatures it can be shown that a reached temperature is stable and has only very little fluctuations. If cycling is carried out, the temperature variation from one cycle to the other is very low, under 0.5 degrees of Celsius, which is satisfactory for the planned applications. For the determination of LOD, the signals are the grey scale values of the image taken by the CCD camera. The result is a mean grey value of 68 whereas the lowest concentration has a value of 124.
The calculated LOD is at a concentration of 0.03 microgram / ml. PCR is executed and results are shown. The lowest detectable concentration using E. coli plasmid was under 100 copies / microlitre. To increase the dynamic range, all measurements use several images taken with different integration times, with this technique, a dynamic range of more than three orders of magnitude is achieved.
Analytical validation of cartridges and reagents
A hydrolysis probe PCR is executed automatically on the instrument, using E. coli plasmid (pTYB1) as template. The concentrations of the template were 1 000 copies / microlitre, 100 copies / microlitre and 10 copies / microlitre in the reaction. Several experiments were executed for reducing the holding time for the PCR steps in the BCR 603. The fastest PCR run was done in about one hour.
Software validation
The function of the software is not only to communicate with the user, but control and execution of the whole process. After starting the software, it is checked if the components are connected and working. The user interface is disabled until this test is done. Error messages will pop-up if one component is not working. Every time a component should be used and could not be detected the software stops at this status and shows an error message. Other tests are done like checking the user inputs for the right syntax. In chips temperature measurement, ramp rate determinations, overshooting measurements were used to validate the software. The gained information was used to improve functionally and performance of the software. During the project the software gradually included the following features: extension of the dynamic range by using different integration times, automatic feature recognition and calculation of PCR curves and analytical values. The software and the interfaced platform were validated by functional tests by comparison with gold standard molecular diagnostic procedures.
WP7 - Clinical validation
In order to design a CIN Point of Care Diagnostic device that is generally accepted among the medical community, it will be necessary to convince the medical and scientific community that the new system is as or more reliable than their current conventional diagnostic methods.
Extension and maintenance of cervical specimen bioresource of clinical specimens (TCD)
Bioresource of cervical smear specimens has been extended and the repertoire of tests performed increased to include CIN genotype analysis. A comprehensive collection of liquid based cervical cytology smear samples and nucleic acid specimens has been collected and material from up to 2 000 patients is available within the CERVIVA biobank at TCD. The total nucleic acid has been banked at -80 degrees of Celsius until required for validation of the AUTOCAST biomarker panel. To date our CERVIVA biobank consists of specimens (ThinPrep and nucleic acid specimens) from over 2 000 women attending for first visits at the colposcopy clinic at the Coombe Women and Infants University Hospital, Dublin, Ireland. The cytological diagnosis and other testing modalities (Hybrid Capture 2, Roche Linear Array CIN Genotyping Test and others) have been made at the baseline. The majority of women have had a colposcopic procedure performed after taking the initial smear specimen. These procedures included biopsy (Punch or LOOP), or LLETS treatment (Large Loop Excision of the Transformation sone), and histological data from these patients is established. The overall output of this biobank is a well characterised comprehensive collection of cervical cytology smear samples and the corresponding nucleic acid extract, with defined cytology and CIN DNA and mRNA status. This biobank of specimens is available for all primary testing, validation, and proof of principle experiments at each stage of the AUTOCAST chip and chemistry development.
Validation of CIN mediator Probe and molecular beacon technology for detection of CIN clinical specimens
Specimens within the resource bank have already been used to evaluate the sensitivity and specificity of the current gold standard CIN detection methods and Genoid background CIN detection technologies. The existing GenoID molecular beacon real time CIN PCR test to the gold standard CIN detection technology hc2 and the full spectrum CIN test from GenoID, furthermore specific CIN genotyping were confirmed by comparison with the reference genotyping test CIN Linear Array. We have demonstrated the real time PCR based approaches to CIN detection for cervical screening are comparable in many aspects to the gold standard CIN test hc2, which uses in vitro nucleic acid hybridisation assay and signal amplification and to the more standard PCR hybridisation assays such as the full spectrum CIN test (FS-CIN) (GenoID) and CIN linear array (Roche). The sensitivity for hc2, FSCIN and MBRT CIN was 98.3 %, 97.4 % and 93.9 % respectively, while the positive predictive value for hc2, FSCIN and MBRT CIN was 94.1 %, 94.1 % and 97.3 % respectively.
The MBRT CIN test appears to have a slightly higher PPV than the other tests. A bank of CIN positive samples (n equal to 241) with full CIN genotype assessment has been performed. As expected CIN 16 was the most prevalent type in this population representing 66.4 % of the CIN positive cases, followed by CIN31 (22.1 %), and CIN33 (18.3 %). The multiple infection rate among the population genotyped was 58 % which is high compared with a routine screening population, but this is explained by the high prevalence of CIN in the selected population.
The evaluation of the liquid phase MPA in comparison with the current gold standard CIN detection technologies was also carried out. The performance of the liquid phase MPAs for detection of CIN 16, 18, 31 and 33 in cell lines and clinical specimens was demonstrated. 1-10 cells per reaction or 210 fg / microlitre of CaSKI DNA, 25 pg / microlitre of SiHA cells sensitivities were demonstrated.. LOD for amplification of CIN18 DNA was comparable, at 78.3 copies per 10 microlitre reaction (95 % confidence interval (CI): 47.0 to 372.5) which was comparable to hydrolysis assay (TaqMan PCR).
A clinical evaluation of the CIN16 liquid phase MPA on 185 cervical smear specimens indicated there was 79.8 % concordance between CIN16 MP assay result and Roche Linear Array result for detection of CIN16 positive samples. Similarly, when compared with the GenoID real time molecular beacon assay for CIN16 / 18 duplex assay the CIN 16 MPA showed concordance of 80.1 %.
The CIN16 MPA showed a high positive predictive value of 94.9 % but a sensitivity that was lower (41.1 %) than the other commercially available CIN detection technologies for detection of histologically confirmed high grade disease. This is due to detection of a single genotype by the CIN16 MPA compared to the other tests which detect a greater number of genotypes. The sensitivity of the MPA can be improved by increasing the number of mediator probes to include the panel of 14 high risk CIN subtypes.
Proof-of-principle validation of novel CIN biochip (prototype system) for detection of CIN in clinical specimens
Three AUTOCAST prototype chip formats were used for the 'proof-of-concept' clinical validation:
- methods for CIN detection;
- assessment of the concordance with standard CIN and biomarker detection approaches and determination of the reproducibility and robustness of MPA and the AUTOCAST real time PCR instrument were carried out. The specimens used for validation of the system were from a range of cytological disease categories broken down as follows; negative 2.3 % (2 / 77), Borderline nuclear abnormalities (BNA) 6.5 % (5 / 77), CIN1 9.1 % (7 / 77), CIN2 27.3 % (21 / 77), and CIN3 54.5 % (42 / 77). Histology results were available for 70 of these women. Overall, 2.8 % of the women had negative histology, 11.4 % of the women had CIN1 and 85.7 % had CIN2+.
- Control material for the initial validation experiments included DNA and cDNA generated from a CIN positive cell line CaSki (CIN 16 and 18 positive squamous cell carcinoma). All samples were screened for CIN initially using hc2 (Qiagen) or the Cobas 4800 CIN test, specific CIN genotyping was then performed using the Linear Array CIN test and the molecular beacon real time PCR assay from GenoID. On chip, 85.5 % (47 / 55) were positive for CIN16. This compares with positivity rates of 96.4 % (53 / 55) for linear array CIN genotyping test, 96.4 % (53 / 55) for hybrid capture 2 and 96.4 % (53 / 55) for the GenoID molecular beacon real-time test for CIN16 / 18. The corresponding MPA for CIN16 performed in liquid phase on the AB7900 instrument, demonstrated 100% positivity for CIN16. The sensitivity and positive predictive value of the CIN16 mediator probe on chip for detection of CIN2+ as a true positive were 92.3 % and 97.3 % respectively. Overall, 24 CIN16 positive patient samples, selected based on their biomarker expression patterns on TaqMan PCR were tested on prototype 2 chips (n equal to 24) using the AUTOCAST real-time PCR microarray analyser. Of the samples tested on chip, 83.3 % (20 / 24) were positive for CIN16 and 87.5 % (14 / 16) [that were positive for all 4 biomarkers by TaqMan PCR on AB7900 instrument) were positive on chip using the real-time PCR Microarray Analyser. The current capacity of the AUTOCAST system in terms of multiplexity was demonstrated by concurrent amplification of CIN16 DNA and cDNA for 4 mRNA biomarkers (Serpin, TP63, KLK8 and RTKN2) in respective channels of the second prototype chip. Analytically, the AUTOCAST system showed good reproducibility between runs and across each chip with LOD for the CIN 16 MPA on chip at 10 cells per reaction. This compares well to the capability of TaqMan PCR given that sensitivity of the mediator probe reaction on chip may currently be limited by epifluorescence detection.
Potential impact:
Global cervical cancer incidence increased from 378,000 (256 000 - 489 000) cases per year in 1980 to 454 000 (318 000 - 620 000) cases per year in 2010, a 0.6 % annual rate of increase. Cervical cancer death rates have been decreasing but the disease still killed 200 000 (139 000 276 000) women in 2010, of whom 46 000 (33 000 - 64 000) were aged 15-49 years in developing countries. It is usually preceded by a long phase of pre-invasive disease called CIN. This pre-invasive phase is characterised microscopically as a range of events progressing from cellular atypia to various grades of dysplasia, including CIN before progression to invasive carcinoma. This precursor phase is generally asymptomatic, and can occur over a long period of 10-20 years.
Human papillomavirus (CIN) is the single most important aetiological agent in the pathogenesis of cervical cancer and pre-cancer. About 15-20 types are associated with cervical cancer, of which CIN16, 18, 31, and 45 accounts for 80 % of cervical cancers. CIN vaccination has now become part of cervical cancer screening and prevention strategies worldwide. There are two prophylactic vaccines currently available which protect against two specific high risk CIN types 16 and 18. With the introduction of CIN vaccination, the landscape of cervical pre-cancer will change over time. While the incidence of abnormal smears and high grade disease will decrease over time, it is anticipated that in the future, the incidence of some type of low grade abnormalities may increase as a consequence. Furthermore, in vaccinated populations, the lower prevalence of disease will directly impact on the performance characteristics (i.e. positive predictive and negative predictive values) of current diagnostic tests.
There are no specific clinical features or symptoms which indicate the presence of CIN. Initial diagnosis is usually made by cytological analysis of a Pap smear specimen. Cytological analysis of PAP smears is the current method employed by cervical screening programmes for the initial diagnosis of abnormality. The result of the PAP smear is what refers the women for additional screening in colposcopy. However it is well recognised that there are limitations to the existing PAP smear screening procedures. These include a significant incidence of false positive and false negative results. Efforts to improve the specificity and sensitivity focus on triage of cytology with CIN testing, which certainly can improve the sensitivity but issues with specificity still remain. As eluded to above in CIN vaccination populations, the rate of abnormal cytology will decrease, this will mean it will be more challenging for cytotechnologists to review cervical cytology samples. Its likely therefore that CIN testing will become the primary screening event in vaccinated populations. There remains a lot of debate over the best method for triaging women who test CIN positive into high and low risk groups. In this context biomarkers will be an important consideration. Currently histology is the gold standard procedure in terms of disease diagnosis. However histological diagnosis of CINs is complicated by a variety of cellular changes associated with inflammation, pregnancy and/or atrophy. These changes may mimic precancerous cervical lesions, thereby making traditional cervical histology approaches, subjective and prone to variability. This is reflected in poor inter-observer agreement between pathologists. In particular, the differential diagnosis between immature squamous metaplasia and CIN1 / 2, or between low-grade (CIN1) and high-grade (CIN2 / 3) lesions, tend to be difficult. The limitations associated with histological diagnosis, impact greatly on clinical trials in the area of cervical cancer pre-cancer biomarker validation, CIN vaccination, and novel therapies, as histology is usually the end point in terms of disease. Accurate grading of CIN lesions is paramount for clinical management of patients because CIN 1 and CIN 2-3 lesions are treated differently. Furthermore, as histological diagnosis of cervical biopsy ultimately determines the decision to treat, combined with the low positive predictive value of the smear test which initially refers women into colposcopy, often results in unnecessary treatments. The estimate of this burden has been assessed in several studies, where up to 20 % of low grade lesions were upgraded and 26 % of high-grade lesions were downgraded after histological review. The issues relating to accurate diagnosis of CIN can result in over- or under-treatment of the lesion. Moreover, in early CIN lesions, the progression rates to high-grade lesions are low (CIN1 and 2 to CIN3 is 9-22 % and to invasive cancer is 1-5 %). This effectively means that many women unlikely to progress are being over-treated and followed up unnecessarily, causing a huge burden on healthcare systems. Collectively, this emphasises the need for specific biomarkers to aid objective CIN lesion grading, to identify true high-grade cervical disease, and to increase the specificity and PPV of disease detection. Recent advances have opened new avenues for the biomarker based cervical cancer screening. The AUTOCAST project has been taken significant step toward these goals.
The project delivered significant results including: a novel cervical cancer screening technology, which has already attracted significant attention, a cheap and transportable point-of-care device (POC device), which participates in several ongoing future developments, a patented real-time PCR technology and several, critical, enabling know-how. The original goal to develop a POC with application of CIN and biomarker assay technologies is partially achieved, we have achieved the full integration of the instrumentation, and without solid phase, the integration of the biomarkers assay technologies and the new real-time PCR detection technology. CIN multiplex amplification is also achieved at 14 high-risk and 6 low-risk genotypes were amplified simultaneously. Several members plan to use the generated foreground for commercial or commercialisation targeting follow-up projects. Genoid, the SME in role of the exploitation of the project results, has already started the commercialisation process.
Successful implementation of mediator probe technology was achieved after investigation of several assay principles. The Mediator hybridises at an immobilised hair-pin oligonucleotide exhibiting a specific Mediator complementary site. The polymerase extends the Mediator whereby the 5'-terminus of hair-pin oligonucleotide is digested and the quencher molecule is cleaved off. The polymerase elongates the primer until the polymerase activity is inhibited by an abasic site (optional) or by steric hindrance due to the hair-pin or internal linker. The polymerase dissociates from the oligonucleotide, the elongated Mediator is available for the next hybridisation event. Assay performance is correlated to immobilisation efficiency and reproducibility of the 'Universal Reporter' molecule. The assay principle has the potential to produce universal PCR coupled arrays for high parallel detection of analyses.
Within the project, AUTOCAST has developed the following:
1. A cervical mRNA expression biomarker set for sensitive and specific detection of CIN2+ lesions in cervical smear specimens which has been validated in a large cohort of colposcopy referred patients with histologically confirmed disease status. The complete clinical validation of the selected biomarkers panel was performed on standard real time PCR instrumentation and yielded excellent results. For the detection of CIN2+ cases the biomarker panel specificity (0.93) and sensitivity (0.89) were better than standard screening methods including high risk HPV Deoxyribonucleic acid (DNA) tests (specificity 0.85 sensitivity 0.88) and / or cytology alone with ASCUS+ cutoff (specificity 0.83 sensitivity 0.70) or with LSIL+ cutoff (specificity 0.97 sensitivity 0.54) respectively.
2. A novel real time PCR detection technology for use in solid phase microarray formats called the MPA has been developed and its proof of principle was demonstrated for both liquid and solid phase chemistries. MPA developed within project combines the enormous multiplex capability of microarrays together with the sensitivity of real-time PCR. Additionally the assay enables the use of a universal microarray whilst any gene could be detected by changing the primers and probes of the PCR.
3. A unique PCR cartridge was designed which includes an optical and fluidic chip, with 8 channels, each channel suitable to multiplex up to 20 gene targets. The material for each chip was selected fulfilling production, thermal, optical, chemical, biomolecule immobilisation and PCR biocompatibility requirements. Chips have been produced by injection moulding and several bonding and coating options have been evaluated.
A low-cost total internal reflection fluorescence and epifluorescent PCR cycler / scanner has been produced which has automatic data handling capabilities. Partial integration of these various technologies was tested on clinical specimens using the liquid phase, MPA and PCR scanner technology and targeting the application of the developed biomarker panel.
Project context and objectives:
Globally cervical cancer incidence rates have increased from 378 000 (256 000 - 489 000) cases per year in 1980 to 454 000 (318 000 - 620 000) cases per year in 2010, a 0.6 % annual rate of increase. Cervical cancer death rates have been decreasing but the disease still killed 200 000 (139 000 - 276 000) women in 2010, of whom 46 000 (33 000 - 64000) were aged 15-49 years in developing countries. It is usually preceded by a long phase of pre-invasive disease called Cervical intraepithelial neoplasia (CIN). This pre-invasive phase is characterised microscopically as a range of events progressing from cellular atypia to various grades of dysplasia, including CIN before progression to invasive carcinoma. This precursor phase is generally asymptomatic, and can occur over a long period of 10-20 years.
CIN is the single most important aetiological agent in the pathogenesis of cervical cancer and pre-cancer. About 15-20 types are associated with cervical cancer, of which CIN16, 18, 31, and 45 accounts for 80 % of cervical cancers. CIN vaccination has now become part of cervical cancer screening and prevention strategies worldwide. There are two prophylactic vaccines currently available which protect against two specific high risk CIN types 16 and 18. With the introduction of CIN vaccination, the landscape of cervical pre-cancer will change over time. While the incidence of abnormal smears and high grade disease will decrease over time, it is anticipated that in the future, the incidence of some type of low grade abnormalities may increase as a consequence. Furthermore, in vaccinated populations, the lower prevalence of disease will directly impact on the performance characteristics (i.e. positive predictive and negative predictive values) of current diagnostic tests.
There are no specific clinical features or symptoms which indicate the presence of CIN. Initial diagnosis is usually made by cytological analysis of a Pap smear specimen. Cytological analysis of PAP smears is the current method employed by cervical screening programmes for the initial diagnosis of abnormality. The result of the PAP smear is what refers the women for additional screening in colposcopy. However it is well recognised that there are limitations to the existing PAP smear screening procedures. These include a significant incidence of false positive and false negative results. Efforts to improve the specificity and sensitivity focus on triage of cytology with CIN testing, which certainly can improve the sensitivity but issues with specificity still remain. As eluded to above in CIN vaccination populations, the rate of abnormal cytology will decrease, this will mean it will be more challenging for cytotechnologists to review cervical cytology samples. It is likely therefore that CIN testing will become the primary screening event in vaccinated populations. There remains a lot of debate over the best method for triaging women who test CIN positive into high and low risk groups. In this context biomarkers will be an important consideration. Currently histology is the gold standard procedure in terms of disease diagnosis. However histological diagnosis of CINs is complicated by a variety of cellular changes associated with inflammation, pregnancy and / or atrophy. These changes may mimic precancerous cervical lesions, thereby making traditional cervical histology approaches, subjective and prone to variability. This is reflected in poor inter-observer agreement between pathologists. In particular, the differential diagnosis between immature squamous metaplasia and CIN1 / 2, or between low-grade (CIN1) and high-grade (CIN2 / 3) lesions, tend to be difficult. The limitations associated with histological diagnosis, impact greatly on clinical trials in the area of cervical cancer pre-cancer biomarker validation, CIN vaccination, and novel therapies, as histology is usually the end point in terms of disease. Accurate grading of CIN lesions is paramount for clinical management of patients because CIN 1 and CIN 2-3 lesions are treated differently. Furthermore, as histological diagnosis of cervical biopsy ultimately determines the decision to treat, combined with the low positive predictive value of the smear test which initially refers women into colposcopy, often results in unnecessary treatments. The estimate of this burden has been assessed in several studies, where up to 20 % of low grade lesions were upgraded and 26 % of high-grade lesions were downgraded after histological review. The issues relating to accurate diagnosis of CIN can result in over- or under-treatment of the lesion. Moreover, in early CIN lesions, the progression rates to high-grade lesions are low (CIN1 and 2 to CIN3 is 9-22 % and to invasive cancer is 1-5 %). This effectively means that many women unlikely to progress are being over-treated and followed up unnecessarily, causing a huge burden on healthcare systems. Collectively, this emphasises the need for specific biomarkers to aid objective CIN lesion grading, to identify true high-grade cervical disease, and to increase the specificity and PPV of disease detection.
Molecular methods (especially nucleic acids based assays) are powerful diagnostic tools for rapid detection of specific gene sequences in clinical samples. However, molecular diagnostic tests in general, such as the multiplex polymerase chain reaction (PCR), DNA sequencing and microarrays, are still labour intensive, time consuming and complex investigations. To overcome these limitations and to enable simpler use of these methods for medical diagnostics, the whole process of molecular tests requires simplification and revision with miniaturisation and automation presenting a great potential for the integration of molecular assays in an easy-to-use and flexible point of care diagnostic system. In this project, we have developed a novel low cost automated real time PCR / probe technology, in a microarray biochip format with corresponding automated detection instrumentation for use as a rapid point of care diagnostic device in both a clinical and laboratory setting.
Cervical cancer and its associated virus, CIN is the model system which has been used to develop this novel automated cancer screening technology. The uniqueness of the current project is the potential of the novel real time PCR technology to detect panels of multiple biomarkers, both viral and host cell markers, within a single test. This will impact enormously on current cervical screening programs particularly in the context of CIN vaccination. Incorporation of these biomarker panels into cervical screening programs has the potential to stratify patients based on their CIN subtype, defining different treatment cohorts, resulting in improved patient management and reduced costs to healthcare providers.
The technology described in the presented approach is based on a microfluidic disposable cartridge enabling real-time PCR amplification and detection. The disposable cartridge is processed by a small integrated low cost thermo-cycler and reader system which can perform the diagnostic test automatically. The diagnostic test relies on a panel of biomarkers, which are detected within the cartridge to allow for multi parameter analysis. The detection is based on the most sensitive Total internal reflection fluorescence (TIRF) method and has been modified to facilitate the the more common epifluorescent technique. By this, the sensitivity and specificity of detection can be improved and there is a potential to generate quantitative results.
The novel real time PCR approach, which incorporates a novel type of MPA, is applicable to a large variety of diagnostic markers and PCR-based tests. This approach enables the use of a single generic microarray for multiple purposes. Thus the high volume fabrication of the disposable cartridge hosting the microarray can be largely facilitated and commercial uptake can be fostered. The developed technology has the potential to eliminate the bottlenecks of current immobilised real-time chemistries such as immobilised molecular beacons. The developed system and assay technology can not only be used in the medical field for cancer research and diagnostics, but will also be applicable for food safety, environmental monitoring or homeland security if a suitable biomarker panel is adopted. The applicability and performance of the developed prototype system in the AUTOCAST project has been demonstrated for detection of HPV (CIN) and mRNA biomarkers in cervical cancer.
With the recent introduction of the Gardasil CIN vaccine (Sanoffi-Pasteur) in Europe, the combined use of cervical screening and a CIN prophylactic vaccination programme heralds a new era in cervical cancer prevention. The diagnostic device developed within the current project will advance and complement international cervical cancer vaccination and screening programs, through its ability to stratify patients based on their CIN and biomarkers status, in a point of care setting. As part of the project a clinical study has been performed on clinical specimens to validate the full system i.e. the new MPA for CIN detection in combination with the microfluidic cartridge technology. The novel system has been benchmarked against the gold standards currently used in the clinical laboratories.
Any new device to be used in primary screening protocols should include an aim to increase sensitivity and specificity and reduce inappropriate high referral rates from a targeted screening population. The currently available tools used in cervical screening (PAP smears, hybrid capture CIN testing) in general have low positive predictive value and variable sensitivity (5). This has hindered widespread extension of CIN screening programmes.
Nevertheless, the combination of cytology and CIN testing significantly improve the accuracy of primary cervical screening. Therefore, there is a need for a sensitive and specific, high-throughput diagnostic device which will facilitate CIN and biomarker population screening. The attractiveness of a microchip device with multiple reporter channels facilitating sub-type CIN analysis and other viral (load) and human biomarkers in a unified platform represents a major step forward in cervical cancer screening and diagnostics. Such a platform using a microarray device has the potential for development in other diagnostic systems (including other cancers), biosafety platforms and food technology. Microarrays are in principle ideally suited as miniaturised assay platforms for such a multi parameter diagnostic device. What is required is the coupling of a multiplex real-time amplification step with detection on an array platform, heretofore not possible. This proposal attempts to develop a real-time PCR based array technology utilising novel mediator probe chemistry. The microchip will use CIN sub-type analysis and other human biomarkers as a proof of concept for this technology approach.
Main objectives
The aim of the AUTOCAST project was to develop and to validate a point of care (POC) diagnostic system for cancer screening, combining the methods of real time PCR and DNA microarrays. The complete system will comprises:
- a combined thermal cycler and optical detection system to perform PCR cycling and microarray detection simultaneously;
- a microfluidic cartridge containing light guiding components to enable Attenuated total reflection (ATR) and the immobilised microarray within a reaction chamber, which is amenable for PCR cycling and optical readout by TIRF ;
- a novel type of assay protocol (the MPA) to enable screening of any nucleic acid by a single universal/ generic microarray ;
- a new combinational approach for cervical cancer screening including both CIN and human gene biomarker testing;
- a clinical feasibility and validation study to benchmark the performance of the new test to existing diagnostic platforms with respect to CIN diagnostics.
Project results:
According to annex I objectives the following were achieved:
- A combined thermal cycler and optical detection system to perform PCR cycling and microarray detection simultaneously
An automated PCR microarray analyser has been developed within WP4 of the project AUTOCAST. The system includes an optical readout based on TIRF and epifluorescence and a temperature control unit for PCR cycling. The whole system is controlled by electronics and software for a fully automated operation. The software has several updated version according to the project requirements and also to include several new features. Instruments were used at GenoID, Budapest and Trinity College Dublin, respectively. The third one remained in Freiburg to be used by IMTEK and Fraunhofer IPM. D4.4 deliverables and the associated tasks are completed. The AUTOCAST real-time PCR microarray analyser BCR 603 was tested and characterised. The systems redesigned later (prototype) for the clinical validation was also characterised. The analyser's specifications have been met according to the specification of deliverable 4.1 and it is used the characterisation methods in deliverables D4.2. D6.1 D6.2 and D6.3 and the associated tasks are completed. The D6.4 was joined with D6.2 because the delayed chip production, the reagent and performance validation of the demonstrator design was no longer feasible, so only the prototype system was validated.
- A microfluidic cartridge containing light guiding components to enable ATR and the immobilised microarray within a reaction chamber, which is amenable for PCR cycling and optical readout by TIRF
For cartridge demonstrator the adequate plastic material has been determined that will withstand the operation conditions. The AUTOCAST Cartridge consists of the optical and the fluidic chip. The optical and fluidic chip had been delayed due to injection moulding problems, but later successfully completed. For cartridge prototype several bonding techniques were used, laser welding and adhesive bonding procedures were evaluated. The compatibility with mass production of the laser welding process makes it a very promising candidate as a final bonding technique; however it was incompatible with oxygen plasma treatment, which in turn was necessary for functional probe immobilisation of real-time PCR probes. Several contingency measures were introduced with limited success (selective coating, different non oxygen plasma dependent immobilisation methods, thermal bonding). Adhesive bonding was partly incompatible with the PCR process, and also it cannot support the production of larger numbers of chips, which are necessary for the clinical evaluation. Thermal bonding was also incompatible with oxygen plasma treatment. With delay, however, the D5.2 D5.3 D5.4 D5.5 and D5.6 deliverables and the associated tasks are completed.
- A novel type of assay protocol (the MPA) to enable screening of any nucleic acid by a single universal/ generic microarray
The whole assay integration of the singleplex MPA is presented. Five different concepts were evaluated to detect a liquid phase target gene spatially resolved on a solid support. All concepts are based on the Mediator Probe cleavage accompanied by the release of the Mediator. The different approaches comprise different enzyme classes or significantly modified oligonucleotides like multiple fluorophores or inversed internal linkage. The presented detection techniques were developed and evaluated in regard of manufacturing aspects. Regarding feasibility the concept consists of an immobilised hair-pin shaped oligonucleotide with fluorophore and quencher moieties ('Universal Reporter'). A prerequisite for intensive testing and optimisation is the immobilisation of the oligonucleotides using a thermostable coating of the chip surface (cooperation partner Austrian Institute of Technology), the sealing of the microfluidic and optical part of the cartridge by laser welding (cooperation partner Jenoptik Polymer Systems) was not fully achieved during the timeframe of the project. The low reproducibility of the results hampered our efforts to fully deliver D3.3. D3.1 is done partly and discontinued (this was a risk aversion contingency route), D3.2 deliverable and the associated tasks are completed.
- A new combinational approach for cervical cancer screening including both CIN and human gene biomarker testing
A new expression biomarker set for cervical cancer screening has been established. The established biomarker geneset fulfils the criteria to complete the WP D2.3 but we had postponed the due date of the D2.3 to investigate some other alternative possibilities, and establish contingency genesets for further use. The D2.3 delivered significantly improved results. For the detection of CIN2+ cases the biomarker panel specificity (0.93 95 % CI 0.90-0.96) and sensitivity [0.89 95%CI 0.86-0.91] were better than the values of HR-CIN test (specificity 0.85 95 % CI 0.81-0.88 sensitivity 0.88 95 % CI 0.85-0.90) or cytology alone with ASCUS+ cutoff (specificity 0.83 95 % CI 0.77-0.88 sensitivity 0.70 95 % CI 0.66-0.73) or with LSIL+ cutoff (specificity 0.97 95 % CI 0.93-0.99 sensitivity 0.54 95 % CI 0.51-0.55). Bayesian point estimates of positive (PPV) and negative (NPV) predictive values for a hypothetical, natural history based, screening population with 1 % of CIN2+ prevalence are([PPV: 0.12 0.05 0.04 0.17; NPV: 0.999 0.998 0.996 0.995) respectively. Favouring high screening specificities to keep screening costs low for special settings an optimal cut-off value for the biomarker panel test was selected. This was achieved at a compromise of sensitivity for CIN2+ (0.50 95 % CI 0.48-0.50) however the point estimate of PPV improved significantly to 0.55 and the corresponding NPV is still high (0.994). Nevertheless this cut-off still provides high sensitivity for the detection of cervical carcinoma (0.83 95 % CI 0.70-0.86). D2.1 D2.2 D2.3 deliverables and the associated tasks are completed.
- A clinical feasibility and validation study to benchmark the performance of the new test to existing diagnostic platforms with respect to CIN diagnostics
The overall output of this deliverable is to provide a biobank which consists of a comprehensive collection of well characterised cervical cytology smear samples collected from a defined population of women with cervical abnormalities. The AUTOCAST real-time PCR microarray analyser, novel MPA chemistry and 3 prototype chip formats have been successfully demonstrated for detection of CIN16 and mRNA biomarkers in cervical smear specimens. Also we have successfully demonstrated a multiplex CIN PCR detection, with simultaneous amplification of 14 high-risk CIN genotypes in a single channel. To access its value a case control study using Hungarian and Irish cervical screening populations was set up. Cases comprised 750 women who developed CIN2+ or worse confirmed by histology, and were stratified by age group (under 29 and over) and disease grade (CIN2, CIN3, and carcinoma). Controls (n equal to 250) were women who identified disease free by cytology and were the same age on average as cases, the controls were also stratified by age. Biomarker reverse transcriptase PCR measurements using routine exfoliated cytological samples, high-risk CIN testing, cytology and/or histology were performed and used to evaluate different protocols of testing including combinations of the tests with different test cut-offs. The Bayesian estimates of critical parameters of test performance were calculated and compared for the different stratified groups. D7.1 D7.2 and D7.3 deliverable and the associated tasks has been delivered and reported on schedule.
Detailed scientific and technological results
Here we present a short summary of the results we have generated for each of the specific tasks across the 7 WPs. Overall we present the development of a new diagnostic cervical cancer in vitro diagnostic device capable of CIN and mRNA biomarker detection in clinical samples.
WP1 - Selection of biomarker panel and device requirements
Preanalytical constrains
Within this task we defined the optimal sampling medium for processed and use with the AUTOCAST system. The sampling type and medium we have chosen is liquid-based cytology smears specimen collected preferably into ThinPrep PreservCyt, solution. This will complement another project (SPE_SAFE) we have running alongside AUTOCAST, funded through a Hungarian national grant which is developing a combined sampling and extraction device for RNA and DNA within the same timeframe as the AUTOCAST project. This device is designed to provide long ambient sample stability. Common exploitation is proposed for both foregrounds.
We have also defined the required sensitivities we require both for signal intensities of fluorescent detection and PCR analytical sensitivities, in both cases the projection is that the prototype AUTOCAST device should be capable of achieving similar sensitivities to conventional liquid phase real-time PCR reactions. The reaction volume which we will use in the system is 5 ul per detection channel, comprising of 4 ul reaction mastermix and 1 ul nucleic acid sample volume. In order to achieve good analytical sensitivities for the whole sample volume the SPE_SAFE is expected to deliver a highly concentrated nucleic acid solution. The current AUTOCAST chip design has 8 channels with 20 immobilisation areas in each channel for the detection probes. The current instrument is capable of reading at 2 wavelengths so the maximum number of the reactions is 20 reaction in one channel, altogether 160 reactions per chip. While the complete multiplex real-time PCR reaction was not demonstrated within the timeframe of the project, the initial proof of concept studies have demonstrated the feasibility of this high level multiplexity approach. . The results have demonstrated mediator probe surface detection, multiplex CIN amplification and sensitive detection of liquid phase biomarker expression. Nevertheless for the proposed CIN multiplex detection (for 20 genotypes), typing and viral load determination (quantitative detection), the established liquid phase mediator probe reaction is capable of working at a multiplicity of 2. The 8 channels are divided between positive control, negative control and duplicated target detection for both biomarker and CIN detection tasks. The plan was for the other wavelength to be used for spotting control. A universal thermal cycling profile has been developed for the MPA. The overall aim is that the cycling time is kept below one hour to deliver a viable point-of-care instrument, this is achievable with this universal setup, however the current multiplex CIN protocol takes a bit longer. The detection dynamic range is large enough to cover several magnitudes and does not limit the overall detection performance. The software is already available for control of detection and thermal cycling, and will deliver background and negative control corrected evaluated amplification results.
Evaluation of required technologies
For benchmarking, development of reference materials and reactions was carried out: Genoid supplied BF and IMTEK with CIN reference materials, for the biomarker reactions Genoid established a TaqMan based reaction format for two-site evaluation of the biomarker genes (TCD and Genoid). IMTEK made reference reactions and material available for Genoid and TCD regarding the mediator probe chemistry including probe designs for both mediator and molecular beacon reactions. Regarding material selection and technological decisions, evaluation of the possible immobilisation chemistries in accordance with the optical, thermal, chemical, biochemical features of the cartridge material was carried out. Physical design requirements were set as described above. To ensure a low cost device, the technologies are carefully designed, low cost, high productivity injection moulding, laser welding, two-step (chemistry and probes are spotted in two consecutive step in the same technology step), top-spot immobilisation was the planned approach.
Selection of biomarker panel
Selection of genes for the biomarker panel was carried out. The established classification biomarker panel of 20 genes were initially validated on an independent set of samples from TCD biobank (136 samples) and at TCD using independent instrument and personnel and later was validated at Genoid on larger population. All of the cases comprised 750 women who developed CIN2+ or worse confirmed by histology, and were stratified by age group (under 29 and over) and disease grade (CIN2, CIN3, and carcinoma). Controls (n equal to 250) were women who identified disease free by cytology and were the same age on average as cases, the controls were also stratified by age. For the detection of CIN2+ cases the biomarker panel specificity (0.93 95 % CI 0.90-0.96) and sensitivity (0.89 95 % CI 0.86-0.91) were better than the values of HR-CIN test (specificity 0.85 95 % CI 0.81-0.88 sensitivity 0.88 95 % CI 0.85-0.90) or cytology alone with ASCUS+ cutoff (specificity 0.83 95 % CI 0.77-0.88 sensitivity 0.70 95 % CI 0.66-0.73) or with LSIL+ cutoff (specificity 0.97 95 % CI 0.93-0.99 sensitivity 0.54 95 % CI 0.51-0.55).
WP3 - Assay chemistry
A novel assay chemistry which allows simultaneous real-time amplification and detection of multiple targets in parallel was developed. This new technology is called MPA. This mediator probe mediated detection enables the fabrication of universal oligonucleotide arrays for simultaneous real-time PCR detection. Several approaches using molecular beacon chemistry in solid and liquid phase were also evaluated as a contingency to reduce the risk, should the MPA concept fail. These were later abandoned once the MPA proof of concept was demonstrated. Small volume real-time liquid phase reaction
Low volume real-time liquid phase reaction were tested for molecular beacon chemistries (using red wavelength excitation/emission), which was comparable to normal volume reaction. The optimal polymerase and buffer pair was selected to achieve effective mediator probe cleavage. Liquid-phase mediator probe cleavage was demonstrated.
Solid phase adaptation of molecular beacon assay
Solid phase molecular beacon chemistry was a contingency measure and was published to be successfully implemented in other PCR reactions, so this was discontinued as a task.
Development of mediator probe chemistry
Successful implementation of mediator probe technology was achieved after investigation of several assay principles. The mediator hybridises at an immobilised hair-pin oligonucleotide exhibiting a specific mediator complementary site. The polymerase extends the mediator whereby the 5-terminus of hair-pin oligonucleotide is digested and the quencher molecule is cleaved off. The polymerase elongates the primer until the polymerase activity is inhibited by an abasic site (optional) or by steric hindrance due to the hair-pin or internal linker. The polymerase dissociates from the oligonucleotide, the elongated Mediator is available for the next hybridisation event. Assay performance is correlated to immobilisation efficiency and reproducibility of the 'Universal Reporter' molecule. Several concentrations of the Universal Reporter were tested. The concentration of the Mediator Probe was optimised to 200 nM. Increasing the concentrations neither improved the sensitivity nor increases the fluorescence signal in the assay. The data show a clear correlation between the fluorescence signal plot and the initial DNA copy number. Target DNA down to 10 copies per reaction can be detected. In approx. 95 % of the controls no signal increase was observed showing the specificity of the demonstrated MPA. The cleavage of the Universal Reporter is indispensable for a successful accomplishment of the MPA, especially on solid phase. This implies several requirements for the substrate material, coating and immobilisation chemistry:
- thermostability;
- selective orientation and immobilisation via the linking group;
- non-interfering fluorescence interaction;
- overall feasibility for the assay implementation.
Proof-of-concept studies for the MPA utilising immobilised 'Universal Reporter' molecules within a functionalised microwell ('global coating') were reported in Delieverable D3.2. To transfer the MPA technology to a polymer cartridge with spatially resolved features a simplified setup is presented. An array comprising a limited set of different oligonucleotides is printed on a polymer slide. Furthermore, the polymer slide is bonded with a structured substrate to a microfluidic cartridge. After bonding (Biofluidix) the cartridge was washed (according to a protocol provided by AIT). After heat-activation of the polymerase and cooling down to the annealing temperature, the Mediator hybridises to its complementary sequence site within the Universal Reporter sequence and is processed by the polymerase. The 5' terminus of the 'Universal Reporter' is degraded resulting in an increased fluorescence level. Due to a well-considered array layout, influences like coating, printing buffer, and applied oligonucleotide concentration on thermostability, selective orientation, quenching effects and assay feasibility can be identified and quantified. The intensities of the immobilised oligonucleotides differ significantly. As the control oligonucleotide is modified with a fluorophore the highest signal intensity is detected. The duplexing capability was demonstrated. Two different Mediator molecules trigger signal generation at a specific 'Universal Reporter'. These findings allow the conclusion that principally higher multiplexing degrees are feasible. However, experiments combining target amplification and solid phase detection were not achieved within the timeframe of the project.
WP4 - Real-time PCR microarray analyser
The TIRF based real-time thermocycler array reader (real-time PCR microarray analyser) has been developed and tested. The instrument is capable of reading microarrayed fluorescent signals in real time. In parallel to the assay development, the instrument has been optimised for the diagnostic microarray readout including on chip PCR and real time detection. The partners required instruments for their work at an early stage within the project. Therefore, two stages of development were planned. In a later project stage, the final version of the instrument (second generation) was built up, taking into consideration the experience with the first generation. The second generation fulfils the criteria for clinical validation.
Definition of requirements
The system specifications were defined as a first step of instruments development. Aspects of fluidics and production techniques were taken into account. The requirements include the data of optical detection such as number of dots (20 per channel and 8 channels), area to be read out (8x10 mm), required Limit of detection (LOD) (comparable to state-of-art scanners), wavelength ranges. The temperature requirements for PCR cycling have been defined to make possible amplifications below one hour time limit to support point-of-care applications. The cartridge handling includes a locking mechanism using drawer design and the user interface is a computer program, which makes possible to manipulate all important experimental parameters.
Definition of test procedures for optical and thermal assessment
The definition of test procedures has been defined regarding spot sises surface, the dynamic range and the optical cross talk. Thermal properties include the accuracy and stability of temperature control. All these parameters were measured during analytical evaluation of the prototypes devices.
Demonstrator instrument
On the basis of the specifications, instruments were realised. Simple and automated way to couple the light in after inserting the cartridge into the instrument has been successfully implemented, special optical coupling structures have been designed, tested and realised in the polymer technology. The coupling efficiency and its variations have been characterised and optimised. Homogeneous illumination of the readout area has been achieved, which is necessary for a comparable detection of fluorescence intensity. A low cost, the sensitive detection has also been realised by very simple means in order to keep fabrication costs of the system extremely low. An optimised high aperture imaging optics including spectral filtering has been developed on the basis of commercially available components. High-speed temperature cycling has also been implemented. The instruments have been developed, CE marked and analytically validated. Due to the delay in the delivery of the final AUTOCAST chip, complete overall validation of the system has not been completed to date.
Prototype instrument
In a later project stage, the final version of the instrument (second generation) was built up, taking into consideration the experience with the first generation. In parallel to the instruments development, the cartridge was developed. Therefore, the first instrument operates with standard glass slides, the second generation works with the final cartridge. The second generation was used for clinical validation and proof of concept studies These instruments are operated by clinical lab personnel, requiring a high degree of user friendliness and robustness. Training and maintenance task are necessary to guarantee a successful validation process. Instruments were used at GenoID, Budapest and Trinity College Dublin, respectively. The third one remained in Freiburg to be used by IMTEK and Fraunhofer IPM. The AUTOCAST real-time PCR microarray analyser BCR 603 was developed, tested and modified. The systems redesigned for the clinical validation was characterised. The analysers specifications have been met according to the specification of D4.1 and it is used the characterisation methods in D4.2. D6.1 D6.2 and D6.3 and the associated tasks are completed. The D6.4 was joined with D6.2 because the delayed chip production, the reagent and performance validation of the demonstrator design was no longer feasible, so only the prototype system was validated.
WP5 - Microfluidic cartridge design
Design and fabrication of the disposable microfluidic cartridge including the microarray for multi parameter analysis has been achieved. Establishing a fabrication technology and supply of cartridges to the other work packages throughout the project were delivered and optimisation of the performance in terms of thermal, fluidic and optical properties were carried out.
Selection and validation of materials
Adequate plastic material has been determined which withstands the later operation conditions. The material has been qualified in functional tests regarding mechanical and optical characteristics (low autofluorescence and birefringence), resistance to chemicals and temperature resistance. Furthermore, the material has to meet the relevant criteria for an economic as well as a steady injection moulding process, e.g. in relation to the shape design (micro fluidic cartridge, micro array immobilisation) and sealing. The material has also tested regarding PCR and immobilisation conditions.
Cartridge design
The cartridge consists of an optical chip and a fluidic chip. Both of them are injection moulded. The optical chip incorporates an optical quality readout area and waveguide structures. The fluidic chip has eight detection channels and to sacrificial channels inlets and outlets. The sacrificial channels are applied to prevent clogging of detection channels due to gluing the components, which method is used before the laser welding is available. Before the PCR, channels must be sealed or closed, this has been established.
Construction and tooling and development of the moulding process
JOPS constructed tools for moulding both the optical and fluidic chips. Reproducible moulding of optical chip has been achieved. Production process optimisation contained the following key aspects:
- improvement of the tool operation;
- adaption of machine and systems engineering for realising the process requirements;
- defining a stable working point and process window for injection moulding and injection compression process.
The following research and work has been performed:
- integration of suitable heating and cooling systems with a minimum installation size and high efficiency for implementing the variotherm processes in the tools;
- taking account of the optical component requirements by maintaining the stiffness of the optical contour inserts;
- evaluation and installation of necessary draft angles in the area of the fluidic channels and coating of the tool inserts for reducing the component grip and guaranteeing a damage free ejection;
- researches for improving the polishing as well as the surface quality of the optical coupling and reflective surfaces;
- evaluation and adjusting the pre-stressing of the frame plates for ensuring the filling and compression operation;
- testing and optimising the overall function of the tools;
- purchasing, installation and testing applicable compression software for machines and hardware adaption;
- selection, purchasing, installation and testing applicable heating and cooling technology for realising the required variotherm process;
- evaluation the influence of the process parameters on the component quality.
The following parameters and machine settings had an impact on the process result: injection rate, injection and holding pressure, change-over-point to holding pressure, melt temperature, mould temperature or mould temperature profile, cooling time, clamping force and clamping force profile, stamping stroke, stamping start and stamping speed, development and qualifying process according QS 9000 and ISO 13485.
Optimised and qualified moulding process
To assure that a forming tool or mould together with the appropriate machine and the production process has the ability to manufacture tool and mould-bounded parts that meets the requirements of the AUTOCAST system, several steps were involved. Basic components of the qualifying process: assignment and release of manufacturing drawings, realisation of the tool or mould, risk assessment, qualification (installation qualification, operation qualification), initial sample testing, short-term process analysis (Ppk-Analysis), and production handover were performed. A statement regarding the achieved quality and stability of the process is possible with a Short-term process analysis (Ppk-Analysis). The chosen processes were evaluated and analysed accordingly. The single measurement values of the mechanical dimensions as well as of the birefringence resulted in very good to satisfactory PpK-values. Based on this, the capability of the chosen and analysed processes was demonstrated.
Chip pre-treatment and immobilisation chemistry
Providing a suitable coating is highly challenging as polymer surfaces behave differently compared to glass surfaces. Further, coatings which feature high immobilisation efficiency must not necessarily be suitable for the solid-phase MPA, especially when looking at specificity or thermal stability. Thus, coatings which require more complex (i.e. multi-step) application procedures which can complicate chip fabrication have been investigated. Finally, this led to the identification of potential coating candidates (termed advanced coatings).
Spotting and validation of the microarray
The printing of the assay-specific reagents on the immobilisation coating must fulfil two requirements. First, the printing concept itself must comply with the boundary conditions of the AUTOCAST cartridge design and allow for rapid chip production. Second, independent of the applied coating, the reagent droplets must be printed accurately into the respective channels. Both requirements are met as the current printing concept and instrumentation (TopSpot D) allows for the printing of up to 250 chips per day while ensuring that the reagent spots stay well inside the channels even if assuming worst case scenarios. The array quality (i.e. fluorescence intensity and S / N-ration) has been determined by conventional microarray hybridisation experiments.
Development of the sealing process
During the project, chips were mostly joined by adhesive bonding (please refer to D5.2) and by this process, strongly bonded chips could be produced with high yields. However, this process did only offer limited throughput thus limiting the number of chips that could be produced for the clinical trials. It was thus decided to develop a joining process based on laser welding which offers excellent scalability due to the short required process time per chip. Finally, a new method was developed which featured the best yields and lead to spotted and sufficiently welded cartridges. In this approach, the optical chip was first covered by a laser cut PSA (pressure sensitive adhesive) foil. Then, the complete coating process chain (i.e. plasma treatment and multiple wet chemical processing steps, please refer to deliverable D5.5) could be applied to these chips. It was demonstrated that this method works in principle and valid joining processes can be established this way, with occasional errors and interference effects to be eliminated.
WP6 - System integration
During the prototyping of the device (CIN biochip) it was essential to monitor the fulfilment of the requirements of the prototype and the final device. Analytical validation of the prototypes included: thermal, optical and fluidic validation, software validation and performance validation / cross validation.
Definition of test procedures and thermal and optical characterisation
Measurement of temperature with sensors was carried out. The specified temperatures should be reached in the reagents inside the channels of the chip. As during the PCR cycling no temperature measurement inside the channel is possible, the sensor for the temperature control is fixed on the surface of the peltier element. With PT100 (and other sensor types) the temperature inside the chip can be reached as specified. However the measured temperatures inside the chip are not reproducible in the specified range. Further improvement of the heat transfer between the instrument and the chip should improve the reproducibility. The ramp rates inside the chip have a peak rate of maximum 4 degrees of Celsius / second and an average of maximum 2 degrees of Celsius / second. To enhance the ramp rates several improvements were tested: the thickness of the chips is reduced, the temperature transfer was optimised and an enquiry for peltier elements with higher power was carried out. Despite the inaccuracy of temperatures it can be shown that a reached temperature is stable and has only very little fluctuations. If cycling is carried out, the temperature variation from one cycle to the other is very low, under 0.5 degrees of Celsius, which is satisfactory for the planned applications. For the determination of LOD, the signals are the grey scale values of the image taken by the CCD camera. The result is a mean grey value of 68 whereas the lowest concentration has a value of 124.
The calculated LOD is at a concentration of 0.03 microgram / ml. PCR is executed and results are shown. The lowest detectable concentration using E. coli plasmid was under 100 copies / microlitre. To increase the dynamic range, all measurements use several images taken with different integration times, with this technique, a dynamic range of more than three orders of magnitude is achieved.
Analytical validation of cartridges and reagents
A hydrolysis probe PCR is executed automatically on the instrument, using E. coli plasmid (pTYB1) as template. The concentrations of the template were 1 000 copies / microlitre, 100 copies / microlitre and 10 copies / microlitre in the reaction. Several experiments were executed for reducing the holding time for the PCR steps in the BCR 603. The fastest PCR run was done in about one hour.
Software validation
The function of the software is not only to communicate with the user, but control and execution of the whole process. After starting the software, it is checked if the components are connected and working. The user interface is disabled until this test is done. Error messages will pop-up if one component is not working. Every time a component should be used and could not be detected the software stops at this status and shows an error message. Other tests are done like checking the user inputs for the right syntax. In chips temperature measurement, ramp rate determinations, overshooting measurements were used to validate the software. The gained information was used to improve functionally and performance of the software. During the project the software gradually included the following features: extension of the dynamic range by using different integration times, automatic feature recognition and calculation of PCR curves and analytical values. The software and the interfaced platform were validated by functional tests by comparison with gold standard molecular diagnostic procedures.
WP7 - Clinical validation
In order to design a CIN Point of Care Diagnostic device that is generally accepted among the medical community, it will be necessary to convince the medical and scientific community that the new system is as or more reliable than their current conventional diagnostic methods.
Extension and maintenance of cervical specimen bioresource of clinical specimens (TCD)
Bioresource of cervical smear specimens has been extended and the repertoire of tests performed increased to include CIN genotype analysis. A comprehensive collection of liquid based cervical cytology smear samples and nucleic acid specimens has been collected and material from up to 2 000 patients is available within the CERVIVA biobank at TCD. The total nucleic acid has been banked at -80 degrees of Celsius until required for validation of the AUTOCAST biomarker panel. To date our CERVIVA biobank consists of specimens (ThinPrep and nucleic acid specimens) from over 2 000 women attending for first visits at the colposcopy clinic at the Coombe Women and Infants University Hospital, Dublin, Ireland. The cytological diagnosis and other testing modalities (Hybrid Capture 2, Roche Linear Array CIN Genotyping Test and others) have been made at the baseline. The majority of women have had a colposcopic procedure performed after taking the initial smear specimen. These procedures included biopsy (Punch or LOOP), or LLETS treatment (Large Loop Excision of the Transformation sone), and histological data from these patients is established. The overall output of this biobank is a well characterised comprehensive collection of cervical cytology smear samples and the corresponding nucleic acid extract, with defined cytology and CIN DNA and mRNA status. This biobank of specimens is available for all primary testing, validation, and proof of principle experiments at each stage of the AUTOCAST chip and chemistry development.
Validation of CIN mediator Probe and molecular beacon technology for detection of CIN clinical specimens
Specimens within the resource bank have already been used to evaluate the sensitivity and specificity of the current gold standard CIN detection methods and Genoid background CIN detection technologies. The existing GenoID molecular beacon real time CIN PCR test to the gold standard CIN detection technology hc2 and the full spectrum CIN test from GenoID, furthermore specific CIN genotyping were confirmed by comparison with the reference genotyping test CIN Linear Array. We have demonstrated the real time PCR based approaches to CIN detection for cervical screening are comparable in many aspects to the gold standard CIN test hc2, which uses in vitro nucleic acid hybridisation assay and signal amplification and to the more standard PCR hybridisation assays such as the full spectrum CIN test (FS-CIN) (GenoID) and CIN linear array (Roche). The sensitivity for hc2, FSCIN and MBRT CIN was 98.3 %, 97.4 % and 93.9 % respectively, while the positive predictive value for hc2, FSCIN and MBRT CIN was 94.1 %, 94.1 % and 97.3 % respectively.
The MBRT CIN test appears to have a slightly higher PPV than the other tests. A bank of CIN positive samples (n equal to 241) with full CIN genotype assessment has been performed. As expected CIN 16 was the most prevalent type in this population representing 66.4 % of the CIN positive cases, followed by CIN31 (22.1 %), and CIN33 (18.3 %). The multiple infection rate among the population genotyped was 58 % which is high compared with a routine screening population, but this is explained by the high prevalence of CIN in the selected population.
The evaluation of the liquid phase MPA in comparison with the current gold standard CIN detection technologies was also carried out. The performance of the liquid phase MPAs for detection of CIN 16, 18, 31 and 33 in cell lines and clinical specimens was demonstrated. 1-10 cells per reaction or 210 fg / microlitre of CaSKI DNA, 25 pg / microlitre of SiHA cells sensitivities were demonstrated.. LOD for amplification of CIN18 DNA was comparable, at 78.3 copies per 10 microlitre reaction (95 % confidence interval (CI): 47.0 to 372.5) which was comparable to hydrolysis assay (TaqMan PCR).
A clinical evaluation of the CIN16 liquid phase MPA on 185 cervical smear specimens indicated there was 79.8 % concordance between CIN16 MP assay result and Roche Linear Array result for detection of CIN16 positive samples. Similarly, when compared with the GenoID real time molecular beacon assay for CIN16 / 18 duplex assay the CIN 16 MPA showed concordance of 80.1 %.
The CIN16 MPA showed a high positive predictive value of 94.9 % but a sensitivity that was lower (41.1 %) than the other commercially available CIN detection technologies for detection of histologically confirmed high grade disease. This is due to detection of a single genotype by the CIN16 MPA compared to the other tests which detect a greater number of genotypes. The sensitivity of the MPA can be improved by increasing the number of mediator probes to include the panel of 14 high risk CIN subtypes.
Proof-of-principle validation of novel CIN biochip (prototype system) for detection of CIN in clinical specimens
Three AUTOCAST prototype chip formats were used for the 'proof-of-concept' clinical validation:
- methods for CIN detection;
- assessment of the concordance with standard CIN and biomarker detection approaches and determination of the reproducibility and robustness of MPA and the AUTOCAST real time PCR instrument were carried out. The specimens used for validation of the system were from a range of cytological disease categories broken down as follows; negative 2.3 % (2 / 77), Borderline nuclear abnormalities (BNA) 6.5 % (5 / 77), CIN1 9.1 % (7 / 77), CIN2 27.3 % (21 / 77), and CIN3 54.5 % (42 / 77). Histology results were available for 70 of these women. Overall, 2.8 % of the women had negative histology, 11.4 % of the women had CIN1 and 85.7 % had CIN2+.
- Control material for the initial validation experiments included DNA and cDNA generated from a CIN positive cell line CaSki (CIN 16 and 18 positive squamous cell carcinoma). All samples were screened for CIN initially using hc2 (Qiagen) or the Cobas 4800 CIN test, specific CIN genotyping was then performed using the Linear Array CIN test and the molecular beacon real time PCR assay from GenoID. On chip, 85.5 % (47 / 55) were positive for CIN16. This compares with positivity rates of 96.4 % (53 / 55) for linear array CIN genotyping test, 96.4 % (53 / 55) for hybrid capture 2 and 96.4 % (53 / 55) for the GenoID molecular beacon real-time test for CIN16 / 18. The corresponding MPA for CIN16 performed in liquid phase on the AB7900 instrument, demonstrated 100% positivity for CIN16. The sensitivity and positive predictive value of the CIN16 mediator probe on chip for detection of CIN2+ as a true positive were 92.3 % and 97.3 % respectively. Overall, 24 CIN16 positive patient samples, selected based on their biomarker expression patterns on TaqMan PCR were tested on prototype 2 chips (n equal to 24) using the AUTOCAST real-time PCR microarray analyser. Of the samples tested on chip, 83.3 % (20 / 24) were positive for CIN16 and 87.5 % (14 / 16) [that were positive for all 4 biomarkers by TaqMan PCR on AB7900 instrument) were positive on chip using the real-time PCR Microarray Analyser. The current capacity of the AUTOCAST system in terms of multiplexity was demonstrated by concurrent amplification of CIN16 DNA and cDNA for 4 mRNA biomarkers (Serpin, TP63, KLK8 and RTKN2) in respective channels of the second prototype chip. Analytically, the AUTOCAST system showed good reproducibility between runs and across each chip with LOD for the CIN 16 MPA on chip at 10 cells per reaction. This compares well to the capability of TaqMan PCR given that sensitivity of the mediator probe reaction on chip may currently be limited by epifluorescence detection.
Potential impact:
Global cervical cancer incidence increased from 378,000 (256 000 - 489 000) cases per year in 1980 to 454 000 (318 000 - 620 000) cases per year in 2010, a 0.6 % annual rate of increase. Cervical cancer death rates have been decreasing but the disease still killed 200 000 (139 000 276 000) women in 2010, of whom 46 000 (33 000 - 64 000) were aged 15-49 years in developing countries. It is usually preceded by a long phase of pre-invasive disease called CIN. This pre-invasive phase is characterised microscopically as a range of events progressing from cellular atypia to various grades of dysplasia, including CIN before progression to invasive carcinoma. This precursor phase is generally asymptomatic, and can occur over a long period of 10-20 years.
Human papillomavirus (CIN) is the single most important aetiological agent in the pathogenesis of cervical cancer and pre-cancer. About 15-20 types are associated with cervical cancer, of which CIN16, 18, 31, and 45 accounts for 80 % of cervical cancers. CIN vaccination has now become part of cervical cancer screening and prevention strategies worldwide. There are two prophylactic vaccines currently available which protect against two specific high risk CIN types 16 and 18. With the introduction of CIN vaccination, the landscape of cervical pre-cancer will change over time. While the incidence of abnormal smears and high grade disease will decrease over time, it is anticipated that in the future, the incidence of some type of low grade abnormalities may increase as a consequence. Furthermore, in vaccinated populations, the lower prevalence of disease will directly impact on the performance characteristics (i.e. positive predictive and negative predictive values) of current diagnostic tests.
There are no specific clinical features or symptoms which indicate the presence of CIN. Initial diagnosis is usually made by cytological analysis of a Pap smear specimen. Cytological analysis of PAP smears is the current method employed by cervical screening programmes for the initial diagnosis of abnormality. The result of the PAP smear is what refers the women for additional screening in colposcopy. However it is well recognised that there are limitations to the existing PAP smear screening procedures. These include a significant incidence of false positive and false negative results. Efforts to improve the specificity and sensitivity focus on triage of cytology with CIN testing, which certainly can improve the sensitivity but issues with specificity still remain. As eluded to above in CIN vaccination populations, the rate of abnormal cytology will decrease, this will mean it will be more challenging for cytotechnologists to review cervical cytology samples. Its likely therefore that CIN testing will become the primary screening event in vaccinated populations. There remains a lot of debate over the best method for triaging women who test CIN positive into high and low risk groups. In this context biomarkers will be an important consideration. Currently histology is the gold standard procedure in terms of disease diagnosis. However histological diagnosis of CINs is complicated by a variety of cellular changes associated with inflammation, pregnancy and/or atrophy. These changes may mimic precancerous cervical lesions, thereby making traditional cervical histology approaches, subjective and prone to variability. This is reflected in poor inter-observer agreement between pathologists. In particular, the differential diagnosis between immature squamous metaplasia and CIN1 / 2, or between low-grade (CIN1) and high-grade (CIN2 / 3) lesions, tend to be difficult. The limitations associated with histological diagnosis, impact greatly on clinical trials in the area of cervical cancer pre-cancer biomarker validation, CIN vaccination, and novel therapies, as histology is usually the end point in terms of disease. Accurate grading of CIN lesions is paramount for clinical management of patients because CIN 1 and CIN 2-3 lesions are treated differently. Furthermore, as histological diagnosis of cervical biopsy ultimately determines the decision to treat, combined with the low positive predictive value of the smear test which initially refers women into colposcopy, often results in unnecessary treatments. The estimate of this burden has been assessed in several studies, where up to 20 % of low grade lesions were upgraded and 26 % of high-grade lesions were downgraded after histological review. The issues relating to accurate diagnosis of CIN can result in over- or under-treatment of the lesion. Moreover, in early CIN lesions, the progression rates to high-grade lesions are low (CIN1 and 2 to CIN3 is 9-22 % and to invasive cancer is 1-5 %). This effectively means that many women unlikely to progress are being over-treated and followed up unnecessarily, causing a huge burden on healthcare systems. Collectively, this emphasises the need for specific biomarkers to aid objective CIN lesion grading, to identify true high-grade cervical disease, and to increase the specificity and PPV of disease detection. Recent advances have opened new avenues for the biomarker based cervical cancer screening. The AUTOCAST project has been taken significant step toward these goals.
The project delivered significant results including: a novel cervical cancer screening technology, which has already attracted significant attention, a cheap and transportable point-of-care device (POC device), which participates in several ongoing future developments, a patented real-time PCR technology and several, critical, enabling know-how. The original goal to develop a POC with application of CIN and biomarker assay technologies is partially achieved, we have achieved the full integration of the instrumentation, and without solid phase, the integration of the biomarkers assay technologies and the new real-time PCR detection technology. CIN multiplex amplification is also achieved at 14 high-risk and 6 low-risk genotypes were amplified simultaneously. Several members plan to use the generated foreground for commercial or commercialisation targeting follow-up projects. Genoid, the SME in role of the exploitation of the project results, has already started the commercialisation process.
Successful implementation of mediator probe technology was achieved after investigation of several assay principles. The Mediator hybridises at an immobilised hair-pin oligonucleotide exhibiting a specific Mediator complementary site. The polymerase extends the Mediator whereby the 5'-terminus of hair-pin oligonucleotide is digested and the quencher molecule is cleaved off. The polymerase elongates the primer until the polymerase activity is inhibited by an abasic site (optional) or by steric hindrance due to the hair-pin or internal linker. The polymerase dissociates from the oligonucleotide, the elongated Mediator is available for the next hybridisation event. Assay performance is correlated to immobilisation efficiency and reproducibility of the 'Universal Reporter' molecule. The assay principle has the potential to produce universal PCR coupled arrays for high parallel detection of analyses.
