A deep learning algorithm to detect cutaneous squamous cell carcinoma on frozen sections in Mohs micrographic surgery: A retrospective assessment.

Intraoperative margin analysis is crucial for the successful removal of cutaneous squamous cell carcinomas (cSCC). Artificial intelligence technologies (AI) have previously demonstrated potential for facilitating rapid and complete tumour removal using intraoperative margin assessment for basal cell carcinoma. However, the varied morphologies of cSCC present challenges for AI margin assessment. The aim of this study was to develop and evaluate the accuracy of an AI algorithm for real-time histologic margin analysis of cSCC. To do this, a retrospective cohort study was conducted using frozen cSCC section slides. These slides were scanned and annotated, delineating benign tissue structures, inflammation and tumour to develop an AI algorithm for real-time margin analysis. A convolutional neural network workflow was used to extract histomorphological features predictive of cSCC. This algorithm demonstrated proof of concept for identifying cSCC with high accuracy, highlighting the potential for integration of AI into the surgical workflow. Incorporation of AI algorithms may improve efficiency and completeness of real-time margin assessment for cSCC removal, particularly in cases of moderately and poorly differentiated tumours/neoplasms. Further algorithmic improvement incorporating surrounding tissue context is necessary to remain sensitive to the unique epidermal landscape of well-differentiated tumours, and to map tumours to their original anatomical position/orientation.

OxyChip embedded with radio-opaque gold nanoparticles for anatomic registration and oximetry in tissues.

PURPOSE: Electron paramagnetic resonance oximetry using the OxyChip as an implantable oxygen sensor can directly and repeatedly measure tissue oxygen levels. A phase I, first-in-human clinical study has established the safety and feasibility of using OxyChip for reliable and repeated measurements of oxygen levels in a variety of tumors and treatment regimens. A limitation in these studies is the inability to easily locate and identify the implanted probes in the tissue, particularly in the long term, thus limiting spatial/anatomical registration of the implant for proper interpretation of the oxygen data. In this study, we have developed and evaluated an enhanced oxygen-sensing probe embedded with gold nanoparticles (GNP), called the OxyChip-GNP, to enable visualization of the sensor using routine clinical imaging modalities. METHODS: In vitro characterization, imaging, and histopathology studies were carried out using tissue phantoms, excised tissues, and in vivo animal models (mice and rats). RESULTS: The results demonstrated a substantial enhancement of ultrasound and CT contrast using the OxyChip-GNP without compromising its electron paramagnetic resonance and oxygen-sensing properties or biocompatibility. CONCLUSIONS: The OxyChips embedded with gold nanoparticles (OxyChip-GNP) can be readily identified in soft tissues using standard clinical imaging modalities such as CT, cone beam-CT, or ultrasound imaging while maintaining its capability to make repeated in vivo measurements of tissue oxygen levels over the long term. This unique capability of the OxyChip-GNP facilitates precisely localized in vivo oxygen measurements in the clinical setting.

Modulation of hypoxia by magnetic nanoparticle hyperthermia to augment therapeutic index.

A hypoxic microenvironment in solid tumors has been known to cause resistance to standard therapies and to increase the malignant potential of tumors. The utilization of magnetic nanoparticle hyperthermia (mNPH) has shown promise in improving therapeutic outcome by (1) killing of hypoxic tumor cells directly and (2) increasing tumor oxygenation and therefore susceptibility to therapies. In this study, the interaction of a hypoxic microenvironment with mNPH efficacy was investigated in a human breast cancer orthotopic xenograft model. Using electron paramagnetic resonance (EPR) to assess in vivo oxygen concentration in tumors repeatedly and non-invasively, we found that mNPH increased tumor pO2 from 3.5 to 68.8 mmHg on average for up to 10 days. Tumors treated once with mNPH showed growth delay. On Transmission Electron Microscopy, magnetic nanoparticles (mNPs) were localized intracellularly in multiple vesicles in the cytoplasm of cells within tumors 48 h after incubation of mNP. In conclusion, mNPH increased tumor oxygenation in vivo and resulted in decreased growth of hypoxic tumors. Future studies will establish tumor pO2-guided multimodal therapies, such as mNPH and radiation, to improve therapeutic efficacy.

Skeletal muscle and glioma oxygenation by carbogen inhalation in rats: a longitudinal study by EPR oximetry using single-probe implantable oxygen sensors.

The feasibility of EPR oximetry using a single-probe implantable oxygen sensor (ImOS) was tested for repeated measurement of pO2 in skeletal muscle and ectopic 9L tumors in rats. The ImOS (50 mm length) were constructed using nickel-chromium alloy wires, with lithium phthalocyanine (LiPc, oximetry probe) crystals loaded in the sensor loop and coated with AF 2400(®) Teflon. These ImOS were implanted into the skeletal muscle in the thigh and subcutaneous 9L tumors. Dynamic changes in tissue pO2 were assessed by EPR oximetry at baseline, during tumor growth, and repeated hyperoxygenation with carbogen breathing. The mean skeletal muscle pO2 of normal rats was stable and significantly increased during carbogen inhalation in experiments repeated for 12 weeks. The 9L tumors were hypoxic with a tissue pO2 of 12.8 ± 6.4 mmHg on day 1; however, the response to carbogen inhalation varied among the animals. A significant increase in the glioma pO2 was observed during carbogen inhalation on day 9 and day 14 only. In summary, EPR oximetry with ImOS allowed direct and longitudinal oxygen measurements in deep muscle tissue and tumors. The heterogeneity of 9L tumors in response to carbogen highlights the need to repeatedly monitor pO2 to confirm tumor oxygenation so that such changes can be taken into account in planning therapies and interpreting results.

Recurrent low-dose chemotherapy to inhibit and oxygenate head and neck tumors.

A lack of strategy to counteract hypoxia (pO2 < 10-15 mmHg) and technique to repeatedly measure tumor pO2 has restricted therapeutic optimization. We report the results obtained with an innovative anti-angiogenic strategy of recurrent low-dose (metronomic) chemotherapy to modulate hypoxia and growth of the Head and Neck tumor xenografts.The FaDu tumors were established in the flank of immune deficient mice and EPR oximetry with lithium phthalocyanine crystals was used to follow the temporal changes in tumor pO2 on treatment with gemcitabine including controls for three weeks. The FaDu tumors were hypoxic with a baseline (pre-treatment) pO2 of 2-8 mmHg. A transient increase in the tumor pO2 was evident on day 3 on treatment with a conventional schedule of gemcitabine (150 mg/kg, d1, d8, d15). No significant change in the tumor pO2 on treatment with metronomic gemcitabine (25 mg/kg on d1, d3, d5 for 3 weeks) was observed. However, tumor pO2 increased significantly on d15-d18 during treatment with a metronomic schedule of 15 mg/kg gemcitabine (d1, d3, d5 for 3 weeks). A modest decrease in the tumor growth was evident on treatment with conventional gemcitabine. Notably, tumor growth was significantly inhibited by metronomic (25 and 15 mg/kg) gemcitabine treatment. The immunohistochemistry (IHC) analyses of the tumor samples indicate a decrease in HIF-1α and TSP-1 on treatment with metronomic gemcitabine.In conclusion, a significant inhibition of tumor growth on treatment with metronomic gemcitabine was observed; however, the increase in pO2 was dose dependent. EPR oximetry can be used to follow the temporal changes in tumor pO2 to identify a therapeutic window on treatment with metronomic chemotherapy for potential combination with radiotherapy.

Quantitative pharmacokinetic and biodistribution studies for fluorescent imaging agents.

Pharmacokinetics and biodistribution studies are essential for characterizing fluorescent agents in vivo. However, few simple methods based on fluorescence imaging are available that account for tissue optical properties and sample volume differences. We describe a method for simultaneously quantifying mean fluorescence intensity of whole blood and homogenized tissues in glass capillary tubes for two fluorescent agents, ABY-029 and IRDye 680LT, using wide-field imaging and tissue-specific calibration curves. All calibration curves demonstrated a high degree of linearity with mean R2 = 0.99 ± 0.01 and RMSE = 0.12 ± 0.04. However, differences between linear regressions indicate that tissue-specific calibration curves are required for accurate concentration recovery. The lower limit of quantification (LLOQ) for all samples tested was determined to be < 0.3 nM for ABY-029 and < 0.4 nM for IRDye 680LT.

Identification of a Suitable Untargeted Agent for the Clinical Translation of ABY-029 Paired-Agent Imaging in Fluorescence-Guided Surgery.

PURPOSE: Non-specific uptake and retention of molecular targeted agents and heterogeneous tissue optical properties diminish the ability to differentiate between tumor and normal tissues using molecular targeted fluorescent agents. Paired-agent imaging (PAI) can increase the diagnostic ability to detect tumor tissue by mitigating these non-specific effects and providing true molecular contrast by co-administration of an untargeted control imaging agent with a targeted agent. This study evaluates the suitability of available clinically translatable untargeted agents for the translation of PAI in fluorescence-guided surgery using an affibody-based targeted imaging agent (ABY-029). EXPERIMENTAL: DESIGN: Three untargeted agents that fluoresce near 700 nm and exhibit good clinical safety profiles (methylene blue, IRDye 700DX, and IRDye 680LT) were tested in combination with the clinically tested IRDye 800CW-labeled anti-epidermal growth factor receptor (EGFR) affibody molecule, ABY-029 (eIND 122,681). Properties of the untargeted agent important for human use and integrity of PAI were tested: (1) plasma protein binding; (2) fluorescence signal linearity in in vitro whole blood dilution; (3) in vivo pharmacokinetic matching to targeted agent in negative control tissue; and (4) in vivo diagnostic accuracy of PAI vs single agent imaging (SAI) of ABY-029 alone in orthotopic oral head and neck squamous cell carcinomas. RESULTS: IRDye 680LT outperformed IRDye 700DX and methylene blue with the highest signal linearity (R2 = 0.9998 ± 0.0002, 0.9995 ± 0.0004, 0.91 ± 0.02, respectively), the highest fluorescence yield in whole blood at 1 µM (104.42 ± 0.05, 103.68 ± 0.09, 101.9 ± 0.2, respectively), and the most closely matched ABY-029 pharmacokinetics in EGFR-negative tissues (binding potential error percentage = 0.31% ± 0.37%, 10.25% ± 1.30%, and 8.10% ± 5.37%, respectively). The diagnostic ability of PAI with ABY-029 and IRDye 680LT outperformed conventional SAI with an area-under-the-receiver-operating-characteristic curve (AUC) value of 0.964 vs. 0.854, and 0.978 vs. 0.925 in the Odyssey scanning system and Pearl wide field imaging system, respectively. CONCLUSION: PAI is a highly promising methodology for increasing detection of tumors in fluorescence-guided surgery. Although not yet clinically approved, IRDye 680LT demonstrates promise as an untargeted agent when paired with ABY-029. The clinical translation of PAI to maximize tumor excision, while minimizing normal tissue removal, could improve both patient survival and life quality.

Improved Discrimination of Tumors with Low and Heterogeneous EGFR Expression in Fluorescence-Guided Surgery Through Paired-Agent Protocols.

PURPOSE: The goal of fluorescence-guided surgery (FGS) in oncology is to improve the surgical therapeutic index by enhancing contrast between cancerous and healthy tissues. However, optimal discrimination between these tissues is complicated by the nonspecific uptake and retention of molecular targeted agents and the variance of fluorescence signal. Paired-agent imaging (PAI) employs co-administration of an untargeted imaging agent with a molecular targeted agent, providing a normalization factor to minimize nonspecific and varied signals. The resulting measured binding potential is quantitative and equivalent to in vivo immunohistochemistry of the target protein. This study demonstrates that PAI improves the accuracy of tumor-to-healthy tissue discrimination compared to single-agent imaging for in vivo FGS. PROCEDURES: PAI using a fluorescent anti-epidermal growth factor receptor (EGFR) affibody molecule (ABY-029, eIND 122,681) with untargeted IRDye 700DX carboxylate was compared to ABY-029 alone in an oral squamous cell carcinoma xenograft mouse model at 3 h after dye administration (n = 30). RESULTS: PAI significantly enhanced tumor discrimination, as compared to ABY-029 alone in low EGFR-expressing tumors and highly heterogeneous populations including multiple cell lines with varying expression (diagnostic accuracy: 0.908 vs. 0.854 and 0.908 vs. 0.822; and ROC curve AUC: 0.963 vs. 0.909 and 0.957 vs. 0.909, respectively) indicating a potential for universal FGS image thresholds to determine surgical margins. In addition, PAI achieved significantly higher diagnostic ability than ABY-029 alone 0.25-5-h post injection and exhibited a stronger correlation to EGFR expression heterogeneity. CONCLUSION: The quantitative receptor delineation of PAI promises to improve the surgical therapeutic index of cancer resection in a clinically relevant timeline.

Rapid and Quantitative Intraoperative Pathology-Assisted Surgery by Paired-Agent Imaging-Derived Confidence Map.

PURPOSE: In nonmetastatic head and neck cancer treatment, surgical margin status is the most important prognosticator of recurrence and patient survival. Fresh frozen sectioning (FFS) of tissue margins is the standard of care for intraoperative margin assessment. However, FFS is time intensive, and its accuracy is not consistent among institutes. Mapping the epidermal growth factor receptor (EGFR) using paired-agent imaging (PAI) has the potential to provide more consistent intraoperative margin assessment in a fraction of the time as FFS. PROCEDURES: PAI was carried out through IV injection of an anti-epidermal growth factor receptor (EGFR) affibody molecule (ABY-029, eIND 122,681) and an untargeted IRDye680LT carboxylate. Imaging was performed on 4 µm frozen sections from three oral squamous cell carcinoma xenograft mouse models (n = 24, 8 samples per cell line). The diagnostic ability and tumor contrast were compared between binding potential, targeted, and untargeted images. Confidence maps were constructed based on group histogram-derived tumor probability curves. Tumor differentiability and contrast by confidence maps were evaluated. RESULTS: PAI outperformed ABY-029 and IRDye 680LT alone, demonstrating the highest individual receiver operating characteristic (ROC) curve area under the curve (PAI AUC: 0.91, 0.90, and 0.79) and contrast-to-noise ratio (PAI CNR: 1, 1.1, and 0.6) for FaDu, Det 562, and A253. PAI confidence maps (PAI CM) maintain high tumor diagnostic ability (PAI CMAUC: 0.91, 0.90, and 0.79) while significantly enhancing tumor contrast (PAI CMCNR: 1.5, 1.3, and 0.8) in FaDu, Det 562, and A253. Additionally, the PAI confidence map allows avascular A253 to be differentiated from a healthy tissue with significantly higher contrast than PAI. Notably, PAI does not require additional staining and therefore significantly reduces the tumor delineation time in a 5 [Formula: see text] 5 mm slice from ~ 35 min to under a minute. CONCLUSION: This study demonstrated that PAI improved tumor detection in frozen sections with high diagnostic accuracy and rapid analysis times. The novel PAI confidence map improved the contrast in vascular tumors and differentiability in avascular tumors. With a larger database, the PAI confidence map promises to standardize fluorescence imaging in intraoperative pathology-assisted surgery (IPAS).

Paired-agent imaging as a rapid en face margin screening method in Mohs micrographic surgery.

Background: Mohs micrographic surgery is a procedure used for non-melanoma skin cancers that has 97-99% cure rates largely owing to 100% margin analysis enabled by en face sectioning with real-time, iterative histologic assessment. However, the technique is limited to small and aggressive tumors in high-risk areas because the histopathological preparation and assessment is very time intensive. To address this, paired-agent imaging (PAI) can be used to rapidly screen excised specimens and identify tumor positive margins for guided and more efficient microscopic evaluation. Methods: A mouse xenograft model of human squamous cell carcinoma (n = 8 mice, 13 tumors) underwent PAI. Targeted (ABY-029, anti-epidermal growth factor receptor (EGFR) affibody molecule) and untargeted (IRDye 680LT carboxylate) imaging agents were simultaneously injected 3-4 h prior to surgical tumor resection. Fluorescence imaging was performed on main, unprocessed excised specimens and en face margins (tissue sections tangential to the deep margin surface). Binding potential (BP) - a quantity proportional to receptor concentration - and targeted fluorescence signal were measured for each, and respective mean and maximum values were analyzed to compare diagnostic ability and contrast. The BP and targeted fluorescence of the main specimen and margin samples were also correlated with EGFR immunohistochemistry (IHC). Results: PAI consistently outperformed targeted fluorescence alone in terms of diagnostic ability and contrast-to-variance ratio (CVR). Mean and maximum measures of BP resulted in 100% accuracy, while mean and maximum targeted fluorescence signal offered 97% and 98% accuracy, respectively. Moreover, maximum BP had the greatest average CVR for both main specimen and margin samples (average 1.7 ± 0.4 times improvement over other measures). Fresh tissue margin imaging improved similarity with EGFR IHC volume estimates compared to main specimen imaging in line profile analysis; and margin BP specifically had the strongest concordance (average 3.6 ± 2.2 times improvement over other measures). Conclusions: PAI was able to reliably distinguish tumor from normal tissue in fresh en face margin samples using the single metric of maximum BP. This demonstrated the potential for PAI to act as a highly sensitive screening tool to eliminate the extra time wasted on real-time pathological assessment of low-risk margins.