Thomas A. Roberts

Fluorescence characterization of clinically-important bacteria

Healthcare-associated infections (HCAI/HAI) represent a substantial threat to patient health during hospitalization and incur billions of dollars additional cost for subsequent treatment. One promising method for the detection of bacterial contamination in a clinical setting before an HAI outbreak occurs is to exploit native fluorescence of cellular molecules for a hand-held, rapid-sweep surveillance instrument. Previous studies have shown fluorescence-based detection to be sensitive and effective for food-borne and environmental microorganisms, and even to be able to distinguish between cell types, but this powerful technique has not yet been deployed on the macroscale for the primary surveillance of contamination in healthcare facilities to prevent HAI. Here we report experimental data for the specification and design of such a fluorescence-based detection instrument. We have characterized the complete fluorescence response of eleven clinically-relevant bacteria by generating excitation-emission matrices (EEMs) over broad wavelength ranges. Furthermore, a number of surfaces and items of equipment commonly present on a ward, and potentially responsible for pathogen transfer, have been analyzed for potential issues of background fluorescence masking the signal from contaminant bacteria. These include bedside handrails, nurse call button, blood pressure cuff and ward computer keyboard, as well as disinfectant cleaning products and microfiber cloth. All examined bacterial strains exhibited a distinctive double-peak fluorescence feature associated with tryptophan with no other cellular fluorophore detected. Thus, this fluorescence survey found that an emission peak of 340nm, from an excitation source at 280nm, was the cellular fluorescence signal to target for detection of bacterial contamination. The majority of materials analysed offer a spectral window through which bacterial contamination could indeed be detected. A few instances were found of potential problems of background fluorescence masking that of bacteria, but in the case of the microfiber cleaning cloth, imaging techniques could morphologically distinguish between stray strands and bacterial contamination.

Novel exomphalos genetic mouse model: The importance of accurate phenotypic classification.

Background Rodent models of abdominal wall defects (AWD) may provide insight into the pathophysiology of these conditions including gut dysfunction in gastroschisis, or pulmonary hypoplasia in exomphalos. Previously, a Scribble mutant mouse model (circletail) was reported to exhibit gastroschisis. We further characterise this AWD in Scribble knockout mice. Method Homozygous Scrib knockout mice were obtained from heterozygote matings. Fetuses were collected at E17.5–18.5 with intact amniotic membranes. Three mutants and two control fetuses were imaged by in amnio micro-MRI. Remaining fetuses were dissected, photographed and gut length/weight measured. Ileal specimens were stained for interstitial cells of Cajal (ICC), imaged using confocal microscopy and ICC quantified. Results 127 fetuses were collected, 15 (12%) exhibited AWD. Microdissection revealed 3 mutants had characteristic exomphalos phenotype with membrane-covered gut/liver herniation into the umbilical cord. A further 12 exhibited extensive AWD, with eviscerated abdominal organs and thin covering membrane (intact or ruptured). Micro-MRI confirmed these phenotypes. Gut was shorter and heavier in AWD group compared to controls but morphology/number of ICC was not different. Discussion The Scribble knockout fetus exhibits exomphalos (intact and ruptured), in contrast to the original published phenotype of gastroschisis. Detailed dissection of fetuses is essential ensuring accurate phenotyping and result reporting.

Non-invasive MRI biomarkers for the early assessment of iron overload in a humanized mouse model of β-thalassemia

β-thalassemia (βT) is a genetic blood disorder causing profound and life threatening anemia. Current clinical management of βT is a lifelong dependence on regular blood transfusions, a consequence of which is systemic iron overload leading to acute heart failure. Recent developments in gene and chelation therapy give hope of better prognosis for patients, but successful translation to clinical practice is hindered by the lack of thorough preclinical testing using representative animal models and clinically relevant quantitative biomarkers. Here we demonstrate a quantitative and non-invasive preclinical Magnetic Resonance Imaging (MRI) platform for the assessment of βT in the γβ0/γβA humanized mouse model of βT. Changes in the quantitative MRI relaxation times as well as severe splenomegaly were observed in the heart, liver and spleen in βT. These data showed high sensitivity to iron overload and a strong relationship between quantitative MRI relaxation times and hepatic iron content. Importantly these changes preceded the onset of iron overload cardiomyopathy, providing an early biomarker of disease progression. This work demonstrates that multiparametric MRI is a powerful tool for the assessment of preclinical βT, providing sensitive and quantitative monitoring of tissue iron sequestration and cardiac dysfunction- parameters essential for the preclinical development of new therapeutics.

Investigating Low-Velocity Fluid Flow in Tumors with Convection-MRI

Several distinct fluid flow phenomena occur in solid tumors, including intravascular blood flow and interstitial convection. Interstitial fluid pressure is often raised in solid tumors, which can limit drug delivery. To probe low-velocity flow in tumors resulting from raised interstitial fluid pressure, we developed a novel MRI technique named convection-MRI, which uses a phase-contrast acquisition with a dual-inversion vascular nulling preparation to separate intra- and extravascular flow. Here, we report the results of experiments in flow phantoms, numerical simulations, and tumor xenograft models to investigate the technical feasibility of convection-MRI. We observed a significant correlation between estimates of effective fluid pressure from convection-MRI with gold-standard, invasive measurements of interstitial fluid pressure in mouse models of human colorectal carcinoma. Our results show how convection-MRI can provide insights into the growth and responsiveness to vascular-targeting therapy in colorectal cancers.Significance: A noninvasive method for measuring low-velocity fluid flow caused by raised fluid pressure can be used to assess changes caused by therapy. Cancer Res; 78(7); 1859-72. ©2018 AACR.

Computational fluid dynamics with imaging of cleared tissue and of in vivo perfusion predicts drug uptake and treatment responses in tumours

Understanding the uptake of a drug by diseased tissue, and the drug’s subsequent spatiotemporal distribution, are central factors in the development of effective targeted therapies. However, the interaction between the pathophysiology of diseased tissue and individual therapeutic agents can be complex, and can vary across tissue types and across subjects. Here, we show that the combination of mathematical modelling, high-resolution optical imaging of intact and optically cleared tumour tissue from animal models, and in vivo imaging of vascular perfusion predicts the heterogeneous uptake, by large tissue samples, of specific therapeutic agents, as well as their spatiotemporal distribution. In particular, by using murine models of colorectal cancer and glioma, we report and validate predictions of steady-state blood flow and intravascular and interstitial fluid pressure in tumours, of the spatially heterogeneous uptake of chelated gadolinium by tumours, and of the effect of a vascular disrupting agent on tumour vasculature.The combination of mathematical modelling of tumour tissue, optical imaging of cleared tumours from animal models, and in vivo imaging of vascular perfusion in tumours predicts the tumour uptake and distribution of specific therapeutic agents.

Determinants of ovarian function after response-adapted therapy in patients with advanced Hodgkin's lymphoma (RATHL): a secondary analysis of a randomised phase 3 trial

Summary Background Adverse effects on reproductive function are a key concern in young women treated with chemotherapy for advanced Hodgkins lymphoma. We aimed to identify risk factors for the extent of ovarian damage in women with Hodgkins lymphoma treated with different chemotherapy regimens to inform accurate advice on options for fertility preservation. Methods We recruited female participants from the randomised phase 3 RATHL trial, aged 18–45 years, based on availability of participants at recruiting sites in the UK. The RATHL trial key inclusion criteria were histologically confirmed classic Hodgkins lymphoma, stage IIB–IV or IIA with adverse features (bulky disease or more than two sites of involvement), no previous treatments, and a performance status of 0–3. As part of RATHL, participants were treated with two cycles of doxorubicin, bleomycin, vinblastine, and dacarbazine (ABVD) or AVD followed by an interim PET-CT scan. Participants who had negative interim scans (PET score of 1 to 3 according to the Lugano classification) were randomly assigned (1:1) by use of minimisation, stratified by interim PET score and study centre, to continue ABVD or AVD for four more cycles. Participants with positive scans (PET score of 4 or 5) were escalated to treatment with bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisolone (BEACOPP-14 or escalated BEACOPP) for four cycles. For the protocol-driven prospective cohort substudy, ovarian function was assessed before treatment, during chemotherapy, and then annually for 3 years by use of serum antimüllerian hormone and follicle-stimulating hormone measurements. The RATHL study is registered with ClinicalTrials.gov, number NCT00678327. Findings Between Dec 13, 2010, and Dec 19, 2012, 67 eligible participants were recruited for this prospective cohort study; 57 had received ABVD or AVD (ABVD-AVD group) and ten BEACOPP-14 or escalated BEACOPP (BEACOPP group). Follow-up was fixed at 3 years. Antimüllerian hormone concentrations decreased during both chemotherapy regimens. At 1 year after chemotherapy, antimüllerian hormone concentrations recovered to a median of 10·5 pmol/L (IQR 4·3–17·3) in the ABVD-AVD group, but little recovery was seen after BEACOPP (median 0·11 pmol/L [0·07–0·20]). Age also affected the extent of ovarian function recovery, with antimüllerian hormone recovery in participants aged 35 years or older in the ABVD-AVD group to 37% (SD 10) of their before treatment concentrations, compared with full recovery to 127% (SD 12) in those younger than 35 years (p<0·0001). Follicle-stimulating hormone recovery to less than 25 IU/L occurred for 95% of women younger than 35 years in the ABVD-AVD group by 2 years and was also dependent on age (hazard ratio 0·49, 95% CI 0·37–0·65; p<0·0001). Interpretation Reduced recovery of ovarian function observed in women older than 35 years treated with ABVD or AVD compared with younger women indicates that treatment could reduce their reproductive lifespan and supports discussion of fertility preservation before treatment. Women treated with BEACOPP should be informed of its potential high gonadotoxicity. These findings warrant further investigation in large, prospective studies with fertility and reproductive lifespan as outcomes. Funding Medical Research Foundation and Cancer Research UK.

Deep learning diffusion fingerprinting to detect brain tumour response to chemotherapy

Artificial neural networks are being widely implemented for a range of different biomedical imaging applications.Convolutional neural networks are by far the most popular type of deep earning architecture,but often require very large datasets for robust training and evaluation We introduce deep learning diffusion fingerprinting (DLDF), which we have used to classifydiffusion-weighted magnetic resonance imaging voxels in a mouse model of glioblastoma (GL261 cell line), both prior to and in response to Temozolomide (TMZ) chemotherapy.We show that, even with limited training, DLDF can automatically segment brain tumours from normal brain, can automatically distinguish between young and older (after 9 days of growth) tumours and that DLDF can detect whether or not a tumour has been treated with chemotherapy.Our results also suggest that DLDF can detect localised changes in the underlying tumour microstructure, which are not evident using conventional measurements of the apparent diffusion coefficient (ADC).Tissue category maps generated by DLDF showed regions containing a mixture of normal brain and tumour cells, and in some cases evidence of tumour invasion across the corpus callosum, which were broadly consistent with histology.In conclusion, DLDF shows the potential for applying deep learning on a pixel-wise level,which reduces the need for vast training datasets and could easily be applied to other multi-dimensional imaging acquisitions Abbreviations ANN artificial neural network CT x-ray computed tomography PET positron emission tomography CNN convolutional neural network HARDI high angular resolution diffusion weighted imaging NODDI neurite orientation dispersion and density imaging VERDICT vascular, extracellular and restricted diffusion for cytometry in tumours DLDF deep learning with diffusion fingerprinting TMZ Temozolomide PFA paraformaldehyde H&E hematoxylin and eosin GFAP glial fibrillary acidic protein

Monitoring the Growth of an Orthotopic Tumour Xenograft Model: Multi-Modal Imaging Assessment with Benchtop MRI (1T), High-Field MRI (9.4T), Ultrasound and Bioluminescence.

Background Research using orthotopic and transgenic models of cancer requires imaging methods to non-invasively quantify tumour burden. As the choice of appropriate imaging modality is wide-ranging, this study aimed to compare low-field (1T) magnetic resonance imaging (MRI), a novel and relatively low-cost system, against established preclinical techniques: bioluminescence imaging (BLI), ultrasound imaging (US), and high-field (9.4T) MRI. Methods A model of colorectal metastasis to the liver was established in eight mice, which were imaged with each modality over four weeks post-implantation. Tumour burden was assessed from manually segmented regions. Results All four imaging systems provided sufficient contrast to detect tumours in all of the mice after two weeks. No significant difference was detected between tumour doubling times estimated by low-field MRI, ultrasound imaging or high-field MRI. A strong correlation was measured between high-field MRI estimates of tumour burden and all the other modalities (p < 0.001, Pearson). Conclusion These results suggest that both low-field MRI and ultrasound imaging are accurate modalities for characterising the growth of preclinical tumour models.

In Amnio MRI of Mouse Embryos

Mouse embryo imaging is conventionally carried out on ex vivo embryos excised from the amniotic sac, omitting vital structures and abnormalities external to the body. Here, we present an in amnio MR imaging methodology in which the mouse embryo is retained in the amniotic sac and demonstrate how important embryonic structures can be visualised in 3D with high spatial resolution (100 µm/px). To illustrate the utility of in amnio imaging, we subsequently apply the technique to examine abnormal mouse embryos with abdominal wall defects. Mouse embryos at E17.5 were imaged and compared, including three normal phenotype embryos, an abnormal embryo with a clear exomphalos defect, and one with a suspected gastroschisis phenotype. Embryos were excised from the mother ensuring the amnion remained intact and stereo microscopy was performed. Embryos were next embedded in agarose for 3D, high resolution MRI on a 9.4T scanner. Identification of the abnormal embryo phenotypes was not possible using stereo microscopy or conventional ex vivo MRI. Using in amnio MRI, we determined that the abnormal embryos had an exomphalos phenotype with varying severities. In amnio MRI is ideally suited to investigate the complex relationship between embryo and amnion, together with screening for other abnormalities located outside of the mouse embryo, providing a valuable complement to histology and existing imaging methods available to the phenotyping community.

Integrated and efficient diffusion-relaxometry using ZEBRA

The emergence of multiparametric diffusion models combining diffusion and relaxometry measurements provides powerful new ways to explore tissue microstructure, with the potential to provide new insights into tissue structure and function. However, their ability to provide rich analyses and the potential for clinical translation critically depends on the availability of efficient, integrated, multi-dimensional acquisitions. We propose a fully integrated sequence simultaneously sampling the acquisition parameter spaces required for T1 and T2* relaxometry and diffusion MRI. Slice-level interleaved diffusion encoding, multiple spin/gradient echoes and slice-shuffling are combined for higher efficiency, sampling flexibility and enhanced internal consistency. In-vivo data was successfully acquired on healthy adult brains. Obtained parametric maps as well as clustering results demonstrate the potential of the technique to provide eloquent data with an acceleration of roughly 20 compared to conventionally used approaches. The proposed integrated acquisition, which we call ZEBRA, offers significant acceleration and flexibility compared to existing diffusion-relaxometry studies, and thus facilitates wider use of these techniques both for research-driven and clinical applications.