Prediction of Surgical Upstaging Risk of Ductal Carcinoma In Situ Using Machine Learning Models.

Objective: The purpose of this study was to build machine learning models to predict surgical upstaging risk of ductal carcinoma in situ (DCIS) to invasive cancer and to compare model performance to eligibility criteria used by the Comparison of Operative versus Monitoring and Endocrine Therapy (COMET) active surveillance trial. Methods: Medical records were retrospectively reviewed of all women with DCIS at core-needle biopsy who underwent surgery from 2007 to 2016 at an academic medical center. Multivariable regression and machine learning models were developed to evaluate upstaging-related features and their performance was compared with that achieved using the COMET trial eligibility criteria. Results: Of 1387 women (mean age, 57 years; range, 27-89 years), the upstaging rate of DCIS was 17% (235/1387). On multivariable analysis, upstaging-associated features were presentation of DCIS as a palpable area of concern, imaging finding of a mass, and nuclear grades 2 or 3 at biopsy (P < 0.05). If COMET trial eligibility criteria were applied to our study cohort, then 496 women (42%, 496/1175) would have been eligible for the trial, with an upstaging rate of 12% (61/496). Of the machine learning models, none had a significantly lower upstaging rate than 12%. However, if using the models to determine eligibility, then a significantly larger proportion of women (56%-87%) would have been eligible for active surveillance. Conclusion: Use of machine learning models to determine eligibility for the COMET trial identified a larger proportion of women eligible for surveillance compared with current eligibility criteria while maintaining similar upstaging rates.

Symptomatic ductal carcinoma in situ (DCIS): Upstaging risk and predictors.

PURPOSE: To estimate the upstaging risk of symptomatic ductal carcinoma in situ (DCIS) to invasive disease and to identify features related to upstaging risk. MATERIALS AND METHODS: This retrospective investigation includes symptomatic women with DCIS at core needle biopsy from January 2007 to December 2016 at a large academic institution. Patient characteristics, findings at imaging, core needle biopsy histopathology results, and final surgical histopathology results were retrieved from the medical records. Using standard statistical tests, patient, imaging, and pathology features were compared between DCIS cases that were upstaged to invasive disease at surgery versus cases that were not upstaged. RESULTS: From 2007 to 2016, fewer than 5% (63/1399) of women diagnosed with DCIS presented with symptoms. Therefore, 63 women (mean age, 51; range, 27-88 years) comprised the study cohort. 84.1% (n = 53) presented with an area of clinical concern, and 15.9% (n = 10) presented with pathologic nipple discharge. The most common finding at mammography was calcifications with or without an associated asymmetry or mass (74.1%, 40/54). The upstaging rate of symptomatic DCIS to invasive disease was 34.9% (22/63). Imaging modality used for biopsy was associated with higher upstaging risk, with cases that were biopsied under MRI guidance accounting for 22.7% of upstaged cases versus 4.9% of non-upstaged cases (p = 0.03). CONCLUSIONS: Women with DCIS uncommonly present with symptoms, and the upstaging rate of symptomatic DCIS is high at nearly 35%. Biopsy modality type of MRI is associated with higher upstaging risk.

Revisiting the NIH Stroke Scale as a screening tool for proximal vessel occlusion: can advanced imaging be targeted in acute stroke?

BACKGROUND AND PURPOSE: Most patients with stroke-like symptoms screened by advanced imaging for proximal occlusion will not have a thrombus accessible by neurointerventional techniques. Development of a sensitive clinical scoring system for rapidly identifying patients with an emergent large vessel occlusion could help target limited resources and reduce exposure to unnecessary imaging. METHODS: This historical cohort study included patients who underwent non-contrast CT and CT angiography in the emergency department for stroke-like symptoms. NIH Stroke Scale (NIHSS) criteria were extended to include resolved symptoms and dichotomized as present or absent. Combinations of NIHSS criteria were considered as tests for proximal occlusion. RESULTS: Proximal cerebral vascular occlusion was present in 19.2% (100/522) of the population and, of these, 13% (13/100) had an NIHSS score of 0. The presence on examination or history of diminished consciousness with inability to answer questions, leg weakness, dysarthria, or gaze deviation had 96% sensitivity and 39% specificity for proximal occlusion. If implemented in this population, the use of CT angiography would have been decreased by 32.4% (169/522 patients) while missing 0.76% with proximal occlusions (4/522). Half of those missed (2/4) would have been identified as large vessel infarcts on non-contrast CT, while the remainder (2/4) were transient ischemic attacks associated with carotid stenosis. CONCLUSIONS: In this cohort, specific NIHSS criteria were highly sensitive for emergent large vessel occlusion and, if validated, may allow for clinical screening prior to advanced imaging with CT angiography.