Faculty Bibliography

Despite improvements in the understanding of cancer causation, much remains unknown regarding the mechanisms by which genomic and non-genomic factors initiate carcinogenesis, drive cell invasion and metastasis, and enable cancer to develop. Technological advances have enabled the analysis of whole genomes, comprising thousands of tumours across populations worldwide, with the aim of identifying mutation signatures associated with particular tumour types. Large collaborative efforts have resulted in the identification and improved understanding of causal factors, and have shed light on new opportunities to prevent cancer. In this new era in cancer genomics, discoveries from studies conducted on an international scale can inform evidence-based strategies in cancer control along the cancer care continuum, from prevention to treatment. In this Review, we present the relevant history and emerging frontiers of cancer genetics and genomics from the perspective of global cancer prevention. We highlight the importance of local context in the adoption of new technologies and emergent evidence, with illustrative examples from worldwide. We emphasize the challenges in implementing important genomic findings in clinical settings with disparate resource availability and present a conceptual framework for the translation of such findings into clinical practice, and evidence-based policies in order to maximize the utility for a population.

Diagnosing cancer earlier can enable timely treatment and optimize outcomes. Worldwide, national cancer control plans increasingly encompass early diagnosis programs for symptomatic patients, commonly comprising awareness campaigns to encourage prompt help-seeking for possible cancer symptoms and health system policies to support prompt diagnostic assessment and access to treatment. By their nature, early diagnosis programs involve complex public health interventions aiming to address unmet health needs by acting on patient, clinical, and system factors. However, there is uncertainty regarding how to optimize the design and evaluation of such interventions. We propose that decisions about early diagnosis programs should consider four interrelated components: first, the conduct of a needs assessment (based on cancer-site-specific statistics) to identify the cancers that may benefit most from early diagnosis in the target population; second, the consideration of symptom epidemiology to inform prioritization within an intervention; third, the identification of factors influencing prompt help-seeking at individual and system level to support the design and evaluation of interventions; and finally, the evaluation of factors influencing the health systems' capacity to promptly assess patients. This conceptual framework can be used by public health researchers and policy makers to identify the greatest evidence gaps and guide the design and evaluation of local early diagnosis programs as part of broader cancer control strategies.

Objective: We sought to prospectively evaluate the feasibility of obtaining genetic testing for at least 1 first- or second-degree family member of a proband known to have actionable germline mutation associated with endometrial and/or ovarian cancer through a coordinated referral system. We also identified barriers to genetic assessment in family members. Here we report initial probands screened and their reasons for declining cascade testing.
Method(s): Patients with a diagnosed pathogenic or suspected pathogenic mutation associated with ovarian and/or endometrial cancer were identified from the gynecologic oncology and genetics clinics. If patients did not consent to the study, their reasons for declining participation were documented. Patients who provided consent were asked to contact their first- and/or second-degree relatives to disclose their genetic testing results and advise them to contact our center for a referral to a genetic counselor. The number of relatives per proband who contacted us for a genetic counseling referral was recorded. In addition to providing the referral, we followed up with relatives to determine whether they attended their genetic counseling appointment, received genetic testing, or took any cancer risk-reducing measures based on their results.
Result(s): This study opened in March 2019. To date, we have screened 71 patients and enrolled 26 (37%). Among the 45 patients who were screened but not enrolled, 48.9% (n = 22) reported that their reason for declining participation in the study was that their family members had already received genetic testing. Other common reasons for declining participation were family members refusing testing (17.8%, n = 8) or no eligible family members (17.8%, n = 8) (Table 1).
Conclusion(s): The majority of probands declined participation in this facilitated cascade testing protocol. The most common reason for lack of participation was family members already having genetic testing or not having eligible family members. Patients who declined participation because family members refused testing could benefit from counseling on how to best to communicate with their relatives. Genetic testing for both patients and their relatives is critical to provision of appropriate cancer screening and prevention services. Knowledge of these barriers is important to further improve cascade testing among family members.
Copyright

Automated Visual Examination (AVE) is a deep learning algorithm that aims to improve the effectiveness of cervical precancer screening, particularly in low- and medium-resource regions. It was trained on data from a large longitudinal study conducted by the National Cancer Institute (NCI) and has been shown to accurately identify cervices with early stages of cervical neoplasia for clinical evaluation and treatment. The algorithm processes images of the uterine cervix taken with a digital camera and alerts the user if the woman is a candidate for further evaluation. This requires that the algorithm be presented with images of the cervix, which is the object of interest, of acceptable quality, i.e., in sharp focus, with good illumination, without shadows or other occlusions, and showing the entire squamo-columnar transformation zone. Our prior work has addressed some of these constraints to help discard images that do not meet these criteria. In this work, we present a novel algorithm that determines that the image contains the cervix to a sufficient extent. Non-cervix or other inadequate images could lead to suboptimal or wrong results. Manual removal of such images is labor intensive and time-consuming, particularly in working with large retrospective collections acquired with inadequate quality control. In this work, we present a novel ensemble deep learning method to identify cervix images and non-cervix images in a smartphone-acquired cervical image dataset. The ensemble method combined the assessment of three deep learning architectures, RetinaNet, Deep SVDD, and a customized CNN (Convolutional Neural Network), each using a different strategy to arrive at its decision, i.e., object detection, one-class classification, and binary classification. We examined the performance of each individual architecture and an ensemble of all three architectures. An average accuracy and F-1 score of 91.6% and 0.890, respectively, were achieved on a separate test dataset consisting of more than 30,000 smartphone-captured images.

PURPOSE/OBJECTIVE:Until human papillomavirus (HPV)-based cervical screening is more affordable and widely available, visual inspection with acetic acid (VIA) is recommended by the WHO for screening in lower-resource settings. Visual inspection will still be required to assess the cervix for women whose screening is positive for high-risk HPV. However, the quality of VIA can vary widely, and it is difficult to maintain a well-trained cadre of providers. We developed a smartphone-enhanced VIA platform (SEVIA) for real-time secure sharing of cervical images for remote supportive supervision, data monitoring, and evaluation. METHODS:We assessed programmatic outcomes so that findings could be translated into routine care in the Tanzania National Cervical Cancer Prevention Program. We compared VIA positivity rates (for HIV-positive and HIV-negative women) before and after implementation. We collected demographic, diagnostic, treatment, and loss-to-follow-up data. RESULTS:From July 2016 to June 2017, 10,545 women were screened using SEVIA at 24 health facilities across 5 regions of Tanzania. In the first 6 months of implementation, screening quality increased significantly from the baseline rate in the prior year, with a well-trained cadre of more than 50 health providers who "graduated" from the supportive-supervision training model. However, losses to follow-up for women referred for further evaluation or to a higher level of care were considerable. CONCLUSION/CONCLUSIONS:The SEVIA platform is a feasible, quality improvement, mobile health intervention that can be integrated into a national cervical screening program. Our model demonstrates potential for scalability. As HPV screening becomes more affordable, the platform can be used for visual assessment of the cervix to determine amenability for same-day ablative therapy and/or as a secondary triage step, if needed.