EGFR mutation prevalence, real-world treatment patterns, and outcomes among patients with resected, early-stage, non-small cell lung cancer in Canada.

OBJECTIVES: The ADAURA trial demonstrated the benefit of adjuvant osimertinib among patients with resected, early-stage, epidermal growth factor receptor-mutated (EGFRm) non-small cell lung cancer (NSCLC). To understand the potential population impact, it is critical to deduce the prevalence, management, and outcomes of this patient population in the real-world setting before use of adjuvant osimertinib. MATERIALS AND METHODS: Using PALEOS (Pan-cAnadian Lung cancEr Observational Study) data (2012-2019), a retrospective, multi-center, observational cohort study was conducted among patients with early-stage (IB-IIIA) resected NSCLC who had not received neoadjuvant therapy. Study outcomes included EGFRm prevalence, treatment patterns, recurrence outcomes, and overall and disease-free survival (OS/DFS). RESULTS: Among patients undergoing reflexive EGFRm testing by a pathologist at time of diagnosis irrespective of disease stage (N = 535), 23 % were EGFRm-positive; 15.9 % had common mutations and 5.6 % had uncommon mutations. Within the EGFRm-positive cohort (N = 156), mean age at diagnosis was 68 years, 65 % of patients were female, and 35 % were of Asian descent. At diagnosis, 48 %, 31 %, and 21 % had stage IB, II, or IIIA disease, respectively; 46 % received adjuvant therapy after resection. Half of patients experienced disease recurrence, typically involving distant sites; central nervous system metastasis varied from 12 % to 15.0 % across disease stages. EGFR tyrosine kinase inhibitors were the most commonly received therapy after first metastatic recurrence. Median OS (DFS) was not reached, 71.2 (22.8) months, and 50.1 (18.0) months among stage IB, II, and IIIA patients. Patients with uncommon EGFRm had a lower probability of survival than those with common EGFRm (2 years: 87 % vs 91 %-94 %; 4 years: 56 % vs 73 %-82 %). CONCLUSION: Approximately-one-quarter of patients with resected, early-stage NSCLC were EGFRm-positive in this study. These patients had high recurrence rates and suboptimal long-term survival after treatment with current therapies. New adjuvant treatments are warranted.

Overcoming the limitations of patch-based learning to detect cancer in whole slide images.

Whole slide images (WSIs) pose unique challenges when training deep learning models. They are very large which makes it necessary to break each image down into smaller patches for analysis, image features have to be extracted at multiple scales in order to capture both detail and context, and extreme class imbalances may exist. Significant progress has been made in the analysis of these images, thanks largely due to the availability of public annotated datasets. We postulate, however, that even if a method scores well on a challenge task, this success may not translate to good performance in a more clinically relevant workflow. Many datasets consist of image patches which may suffer from data curation bias; other datasets are only labelled at the whole slide level and the lack of annotations across an image may mask erroneous local predictions so long as the final decision is correct. In this paper, we outline the differences between patch or slide-level classification versus methods that need to localize or segment cancer accurately across the whole slide, and we experimentally verify that best practices differ in both cases. We apply a binary cancer detection network on post neoadjuvant therapy breast cancer WSIs to find the tumor bed outlining the extent of cancer, a task which requires sensitivity and precision across the whole slide. We extensively study multiple design choices and their effects on the outcome, including architectures and augmentations. We propose a negative data sampling strategy, which drastically reduces the false positive rate (25% of false positives versus 62.5%) and improves each metric pertinent to our problem, with a 53% reduction in the error of tumor extent. Our results indicate classification performances of image patches versus WSIs are inversely related when the same negative data sampling strategy is used. Specifically, injection of negatives into training data for image patch classification degrades the performance, whereas the performance is improved for slide and pixel-level WSI classification tasks. Furthermore, we find applying extensive augmentations helps more in WSI-based tasks compared to patch-level image classification.

Chronic Traumatic Encephalopathy-Like Abnormalities in a Routine Neuropathology Service.

Chronic traumatic encephalopathy (CTE) has been described mainly in professional athletes and military personnel and is characterized by deposition of hyperphosphorylated tau at the depths of cortical sulci and around blood vessels. To assess CTE-like changes in a routine neuropathology service, we prospectively examined 111 brains (age 18-60 years). The presence of tau-immunoreactive deposits was staged using guidelines described by others and was correlated with the medical history. 72/111 cases were negative for CTE-like changes; 34/111 were CTE stage <1; 3/111 were CTE stage 1; and 2/111 were CTE stage 2. The combined history of head injury and alcohol and/or drug abuse was a significant predictor of any CTE-like changes. Age was also a significant predictor; most with any CTE-like changes were >40 years old. CTE-like changes were not identified at sites of contusion. Among a separate group studied retrospectively, we identified 4 cases that met full criteria for CTE. We conclude that CTE-like findings are not confined to professional athletes; the risk factors of head injury and substance abuse are similar in the routine population. However, the significance of very small hyperphosphorylated tau deposits remains to be determined. In addition, the absence of typical CTE-like deposits near contusion sites keeps open the question of pathogenesis.

Adenoma and carcinoma components in colonic tumors show discordance for KRAS mutation.

Activating mutations in KRAS are common events in the pathogenesis of colorectal carcinoma and predict response to treatment with anti-EGFR antibodies. Molecular pathology testing for KRAS mutations has become the standard of practice for patients with metastatic colorectal carcinoma. Despite the known histologic and molecular differences between adenomas and carcinomas, the concordance of KRAS mutation between adenomas and carcinomas has not been established leaving some open questions regarding the appropriate choice of tissue for KRAS mutation analysis and correct interpretation of the test results. To address these questions, we analyzed the concordance of KRAS mutation in 70 tumors that contained both adenoma and carcinoma components (2 cases of intramucosal carcinoma, 66 cases with invasion of the submucosa, and 2 invading the muscularis propria). For each case, DNA was separately isolated from the adenoma and the carcinoma component and analyzed for KRAS mutation using direct sequencing. Overall, 30 (43%) of the adenoma cases and 36 (51%) of the carcinoma cases were positive for KRAS mutation. Of the 70 cases, 16 (23%) showed discordant results. Interestingly, the fraction of discordant cases went down as the depth of carcinoma invasion increased. In summary, we identified significant KRAS mutation discordance between the adenoma and carcinoma component of the lesion. Our results suggest that effort should be made to analyze only the invasive component of the lesion and that caution should be taken when interpreting a result based on DNA extracted from noninvasive elements.

Computerized morphometry as an aid in distinguishing recurrent versus nonrecurrent meningiomas.

OBJECTIVE: To use novel digital and morphometric methods to identify variables able to better predict the recurrence of intracranial meningiomas. STUDY DESIGN: Histologic images from 30 previously diagnosed meningioma tumors that recurred over 10 years of follow-up were consecutively selected from the Rambam Pathology Archives. Images were captured and morphometrically analyzed. Novel algorithms of digital pattern recognition using Fourier transformation and fractal and nuclear texture analyses were applied to evaluate the overall growth pattern complexity of the tumors, as well as the chromatin texture of individual tumor nuclei. The extracted parameters were then correlated with patient prognosis. RESULTS: Kaplan-Meier analyses revealed statistically significant associations between tumor morphometric parameters and recurrence times. Tumors with less nuclear orientation, more nuclear density, higher fractal dimension, and less regular chromatin textures tended to recur faster than those with a higher degree of nuclear order, less pattern complexity, lower density, and more homogeneous chromatin nuclear textures (p < 0.01). CONCLUSION: To our knowledge, these digital morphometric methods were used for the first time to accurately predict tumor recurrence in patients with intracranial meningiomas. The use of these methods may bring additional valuable information to the clinician regarding the optimal management of these patients.