Choudhery, Sadia

Abstract Background: DNIV accounts for 6%–10% of newly diagnosed breast cancer cases. Despite widespread mammography screening, its incidence is increasing in the United States and survival of this disease has only modestly improved since the late 1970s. As patient data accumulates in EHR, it’s promising to generate practice-based evidence through utilization of observational data sources. However, assembly of a DNIV cohort based on EHR data is challenging, as key pathologic and staging information are stored in unstructured clinical narratives and not available as structured data. In this study, we developed a rule-based algorithm to phenotype DNIV using natural language processing (NLP) techniques, and implemented the algorithm on our institutional EHR to extract potential DNIV cases. Methods and Results: We defined DNIV as those with either (1) M1 disease identified at time of initial presentation or M1 disease identified within 4 months after definitive surgery. We first developed a reference case list of DNIV verified by physician chart review of the EHR. We next refined the algorithm on a training dataset containing 51 positive and 38 negative reference cases. Next we tested the performance on the testing data containing 23 positive and 55 negative cases. The phenotyping algorithm identified key data elements using NLP, i.e., stage IV breast cancer, definitive surgery, stage 0-III, recurrent breast cancer and associated dates. To identify DNIV cases, phenotyping algorithm integrated temporal relations among the key data elements. The following steps were conducted in the following sequential order: (1) Identification of patients with breast cancer diagnosis using ICD-9 and ICD-10 codes. (2) Patients are positive cases if there are explicit mentions that delineate DNIV from recurrent metastatic breast cancer, such as “de novo stage IV” or “primary intact” detected by NLP at time of diagnosis. (3) Otherwise, patients with definitive surgery are selected if stage IV was within 5 years before definitive surgery or within 4 months after definitive surgery, along with patients with stage IV but without definitive surgery. (4) We further excluded patients with stage 0-III detected within 5 years after stage IV. (5) We further excluded patients with recurrent breast cancer detected before or at time of detection of stage IV. The remaining patients were left as positive cases. Precision of the algorithm was 70%, recall was 87% and F1, the weighted average of precision and recall, was 77%. We implemented our algorithm to interrogate 10 million clinical documents in a cohort of 56,548 patients with breast cancer diagnosis codes who presented to our institution between 2004 and 2018 and had research authorization. We identified 1918 potential DNIV cases. Conclusion: Our future focus is on algorithm refinement. An algorithm-generated cohort could serve as a data source for further study on outcomes related to DNIV and ideally as automated data abstractor and staging process. Citation Format: Liwei Wang, Karthik Giridhar, Kimberly Corbin, Brenda Ernst, Sadia Choudhery, Emmanuel Gabriel, Feichen Shen, Hongfang Liu, James Jakub. Identifying de novo stage IV breast cancer (DNIV) cases in Electronic Health Records (EHR) using natural language processing [abstract]. In: Proceedings of the AACR Virtual Special Conference on Artificial Intelligence, Diagnosis, and Imaging; 2021 Jan 13-14. Philadelphia (PA): AACR; Clin Cancer Res 2021;27(5_Suppl):Abstract nr PO-050.

Time-drive activity-based costing (TDABC) is a practical way of calculating costs, decreasing waste, and improving efficiency. Although TDABC has been utilized in other service industries for years, it has only recently gained attention in healthcare. In this review article, we define the basic concepts and steps of TDABC and provide examples for applications in breast imaging.

Abstract Objective The purpose of this study is to summarize the results of a survey distributed by the Society of Breast Imaging (SBI) to assess applicants’ experience with the 2018 and 2019 Breast Imaging Fellowship Match process. Methods In this institutional review board–exempt study, the SBI issued an anonymous survey to all matched applicants in an attempt to gauge their experience with the 2018 and 2019 Match process. Results The survey was sent to all 2018 and 2019 matched applicants and 105/236 (45%) responses were received. The majority (75%, 79/105) of respondents reported a positive experience with the Match, with at least a 4/5 rating, and only 3% (5/105) reported a rating below 3/5. There was some improvement in 2019, with 86% (24/28) of respondents reporting at least a 4/5 rating compared to 71% (55/77) in 2018. No respondent reported a score below a 3/5 rating in 2019. The most commonly cited issues with the Match were the timing of the Match process, the need for a universal application, and the burden of travel. Location and program reputation were the two most important factors contributing to the final rank order of programs. Conclusion The great majority of applicants felt that the Match created a positive experience. Planned areas of improvement include the implementation of a universal application, the transition to virtual interviews, and a condensed timeline. These measures are likely to increase applicant satisfaction and provide a fair and efficient experience for future breast radiologists.

As tomosynthesis is rapidly adopted by breast imaging practices, tomosynthesis-guided procedures are increasingly being performed. Tomosynthesis-guided needle localizations are feasible and efficient and allow for localization of tomosynthesis-only findings or one-view findings, which may be difficult to localize under standard digital mammography. In this review, we describe our step-by-step approach for performing tomosynthesis-guided localizations of the breast and axilla using a standard tomosynthesis unit.

Abstract Background: In an adaptive randomized trial, when new treatment combinations are being tested, it is important to be able to identify patients who are progressing on treatment so that they can be changed to a different therapeutic regimen. We know that even within the molecularly high risk patients in I-SPY 2, there is considerable variation in biology. In this study, we will present results of using MRI-calculated functional tumor volume (FTV) to identify tumor progression for each breast cancer subtype. Methods: Patients (n=990) enrolled in the I-SPY 2 TRIAL who were randomized to the graduated experimental drug arms or controls from 2010 to 2016 were analyzed. Four MRI exams were performed for each patient: pre-NAC (T0), after 3 weeks of NAC (T1), between regimens (T2), and post-NAC (T3). Functional tumor volume (FTV) was calculated at each exam by summing voxels meeting enhancement thresholds. Tumor progression at T1, T2 or T3 was identified by a positive FTV change relative to T0. Visual inspection was used to exclude false progression due to strong background parenchymal enhancement post-contrast, prominent vessels, motion, or insufficient image quality. pCR was defined as no invasive disease in the breast and lymph nodes. Negative predictive value for pCR was defined as:NPV=number of true non-pCRs / number of patients with MRI assessed tumor progressions, where “true non-pCRs” referred to patients who were non-pCRs at surgery and were assessed as progressors by MRI. The analysis was performed in the full cohort and in sub-cohorts defined by HR and HER2 statuses. Results: Out of 990 patients, 878 had pCR outcome data (pCR or non-pCR, pCR rate = 35%). Total and non-pCR numbers for each subtype, number of patients with tumor progression assessed by MRI at T1, T2, and T3, and NPVs, are shown in Table 1. In the full cohort, the NPV increased consistently over treatment, from T1 (NPV=83%) to T2 (93%), and to T3 (100%). The HER2+ cancer subtypes showed fewer MRI-assessed tumor progressions than HER2- subtypes: e.g. 10/209 (5%) vs. 108/669 (16%) at T1. NPV was 100% for HER2+ subtypes at T1 and T2 except for a single misclassification of a HR- tumor at T1. Only 6 tumor progressors, all HER2- were identified at T3, and all were confirmed at surgery as non-pCRs (NPV=100%). For HR+/HER2-, the NPV increased slightly from 89% at T1 to 91% at T2, while triple negative subtype had a more substantial increase, from 78% to 92%. Conclusions: Our study showed strong association between tumor progressors assessed by MRI with true non-pCRs after NAC. For HER2+ tumors, although MRI progressors are rare, they strongly indicate non-pCR at all treatment time points, while HER2- subtypes show more accurate results later in treatment. We are evaluating MRI change at 6 weeks to determine if that time point is sufficient to predict progressors. Table 1 MRI assessed tumor progression at different treatment time pointN/non-pCRs/%non-pCRMRI assessed tumor progressionT1 (after 3 weeks)T2 (inter-regimen)T3 (post-NAC)NNPV (%)NNPV (%)NNPV (%)Full cohort878/572/65%11883.14192.76100%HR+/HER2-344/280/81%4588.91190.93100%HR+/HER2+134/85/63%610021000N/AHR-/HER2+75/23/31%47521000N/Atriple negative325/184/57%6377.82692.33100% Citation Format: Wen Li, Natsuko Onishi, David C Newitt, Jessica Gibbs, Lisa J Wilmes, Ella F Jones, Bonnie N Joe, Laura S Sit, Christina Yau, A. Jo Chien, Elissa Price, Kathy S Albain, Theresa Kuritza, Kevin Morley, Judy C Boughey, Kathy Brandt, Sadia Choudhery, Amy S Clark, Mark Rosen, Elizabeth S McDonald, Anthony D Elias, Dulcy Wolverton, Kelly Fountain, David M Euhus, Heather S Han, Bethany Niell, Jennifer Drukteinis, Julie E Lang, Janice Lu, Jane L Meisel, Zaha Mitri, Rita Nanda, Donald W Northfelt, Tara Sanft, Erica Stringer-Reasor, Rebecca K Viscusi, Anne M Wallace, Douglas Yee, Rachel Yung, Smita M Asare, Michelle E Melisko, Jane Perlmutter, Hope S Rugo, Richard Schwab, W. Fraser Symmans, Laura J van't Veer, Donald A Berry, Angela DeMichele, Hiroyuki Abe, Deepa Sheth, Kirsten K Edmiston, Erin D Ellis, Richard Ha, Ralph Wynn, Erin P Crane, Charlotte Dillis, Michael Nelson, An Church, Claudine Isaacs, Qamar J Khan, Karen Y Oh, Neda Jafarian, Dae Hee Bang, Christiane Mullins, Stefanie Woodard, Kathryn W Zamora, Haydee Ojeda-Fornier, Pulin Sheth, Linda Hovanessian-Larsen, Mohammad Eghtedari, Michael Spektor, Marina Giurescu, Mary S Newell, Michael A Cohen, Elise Berman, Constance Lehman, William Smith, Kim Fitzpatrick, Marisa H Borders, Wei Yang, Basak Dogan, Sally Goudreau, Thelma Brown, Laura J Esserman, Nola M Hylton. Breast cancer subtype specific association of pCR with MRI assessed tumor volume progression during NAC in the I-SPY 2 trial [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr PD9-04.

Abstract Background: Strong background parenchymal enhancement (BPE) may cause overestimation in tumor volume measured from dynamic contrast-enhanced (DCE) MRI, which may adversely affect the ability of MR tumor volume to predict treatment outcome for patients undergoing neoadjuvant chemotherapy (NAC). Specifically, an overestimation of tumor volume can result in misclassification of patients with complete pathologic response (pCR) as non-responders, leading to less confidence in MRI prediction. As well, overestimation of extent of disease might lead to more aggressive surgical therapy than necessary. This study investigated whether high BPE in the contralateral breast influences the predictive performance of MRI-measured functional tumor volume (FTV) for patients with locally advanced breast cancer undergoing NAC. Methods: patients (n=990) enrolled in the I-SPY 2 TRIAL who were randomized to the graduated experimental drug arms or controls from 2010 to 2016 were analyzed. Each patient had 4 MRI exams: pre-NAC (T0), after 3 weeks of NAC (T1), between NAC regimens (T2), and post-NAC (T3). FTV was calculated at each MRI exam by summing voxels meeting enhancement thresholds. Background parenchymal enhancement (BPE) in the contralateral breast was calculated automatically as mean percentage enhancement on the early (nominal 150sec post-contrast) image in the fibroglandular tissue segmented from 5 continuous axial slices centered in the inferior-to-superior stack. For each treatment time point, patients having both FTV and BPE measurements were included in the analysis. The area under the ROC curve (AUC) was estimated as the association between FTV and pCR at T1, T2, and T3. The analysis was conducted in the full patient cohort and in sub-cohorts defined by hormone receptor (HR) and HER2 status. In each patient cohort, a cut-off BPE value was selected to classify patients with high vs. low BPE by testing AUCs estimated with low-BPE patients reached maximum when the cut-off value varied from median to maximum in steps of 10%. Results: Out of 990 patients, 878 had pCR outcome data (pCR or non-pCR, pCR rate = 35%). Table 1 shows the number of patients, pCR rate, and AUC of FTV for predicting pCR using all patients available vs. a subset patients with low BPE (< BPE cut-off). In the full cohort, AUC increased slightly across all time points after patients with high BPE were removed. In the HR+/HER2- subtype, AUC increased at T1 after removal of cases with high BPE (0.65 vs. 0.71). For HR-/HER2+, AUC increased substantially after removal of high BPE cases (0.65 to 0.86 at T1, 0.71 to 0.87 at T2, and 0.71 to 0.89 at T3), with greater improvement at the early time point (T1) compared to later time points (T2 and T3). Only a slight improvement in the AUC was observed in the HR+/HER2+ and HR-/HER2- subtypes across all time points. Conclusions: High background parenchymal enhancement adversely affected the predictive performance of functional tumor volume measured by DCE-MRI, at early treatment time point for HR+/HER2- and across all time points for HR-/HER2+ cancer subtype. The adverse effect might be offset using subtype-optimized enhancement threshold in calculating functional tumor volume. Table 1 Effect of BPE on the prediction of pCR using FTV at various treatment time pointsT1T2T3npCR rateAUCBPE cut-offnpCR rateAUCBPE cut-offnpCR rateAUCBPE cut-offFullAll64734%0.662762334%0.701761134%0.6925Subset45334%0.6831133%0.7230534%0.72HR+/HER2-All26218%0.651924918%0.718225518%0.7519Subset13118%0.7124818%0.7120419%0.76HR+/HER2+All10636%0.642110538%0.62269634%0.7120Subset5332%0.668438%0.665740%0.73HR-/HER2+All5175%0.65204774%0.71204973%0.7116Subset3073%0.862871%0.872475%0.89HR-/HER2-All22842%0.682822243%0.751821143%0.6916Subset15940%0.7111137%0.7810540%0.75 Citation Format: Wen Li, Natsuko Onishi, David C Newitt, Roy Harnish, Ella F Jones, Lisa J Wilmes, Jessica Gibbs, Elissa Price, Bonnie N Joe, A. Jo Chien, Donald A Berry, Judy C Boughey, Kathy S Albain, Amy S Clark, Kirsten K Edmiston, Anthony D Elias, Erin D Ellis, David M Euhus, Heather S Han, Claudine Isaacs, Qamar J Khan, Julie E Lang, Janice Lu, Jane L Meisel, Zaha Mitri, Rita Nanda, Donald W Northfelt, Tara Sanft, Erica Stringer-Reasor, Rebecca K Viscusi, Anne M Wallace, Douglas Yee, Rachel Yung, Michelle E Melisko, Jane Perlmutter, Hope S Rugo, Richard Schwab, W. Fraser Symmans, Laura J van't Veer, Christina Yau, Smita M Asare, Angela DeMichele, Sally Goudreau, Hiroyuki Abe, Deepa Sheth, Dulcy Wolverton, Kelly Fountain, Richard Ha, Ralph Wynn, Erin P Crane, Charlotte Dillis, Theresa Kuritza, Kevin Morley, Michael Nelson, An Church, Bethany Niell, Jennifer Drukteinis, Karen Y Oh, Neda Jafarian, Kathy Brandt, Sadia Choudhery, Dae Hee Bang, Christiane Mullins, Stefanie Woodard, Kathryn W Zamora, Haydee Ojeda-Fornier, Mohammad Eghedari, Pulin Sheth, Linda Hovanessian-Larsen, Mark Rosen, Elizabeth S McDonald, Michael Spektor, Marina Giurescu, Mary S Newell, Michael A Cohen, Elise Berman, Constance Lehman, William Smith, Kim Fitzpatrick, Marisa H Borders, Wei Yang, Basak Dogan, Laura J Esserman, Nola M Hylton. The effect of background parenchymal enhancement on the predictive performance of functional tumor volume measured in MRI [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P6-02-01.

Abstract Background: The clinical utility of routine annual mammogram after curative-intent treatment for male breast cancer is uncertain. There is potentially greater value after lumpectomy (as surveillance for ipsilateral recurrence) than mastectomy (as screening for a new contralateral cancer). The goal of this study was to assess real world use of mammography in men during the first year after lumpectomy or mastectomy to treat breast cancer. Methods: Administrative claims data from OptumLabs Data Warehouse (a large US database that includes privately insured patients and Medicare Advantage-insured enrollees from all 50 states and of all ages and ethnic and racial groups) were used to identify men treated with breast surgery for a new diagnosis of breast cancer between 2007 and 2017. We required continuous coverage starting at least 6 months prior to the non-metastatic breast cancer diagnosis and continuing until at least 13 months after the breast surgery. Our primary endpoint was the proportion of patients who had at least one mammogram during the year (13-month period, to allow for scheduling and other logistical delays) after lumpectomy or mastectomy. Univariate and multivariate testing were performed to identify predictors of mammography (with p<0.05 used as the threshold for statistical significance for both). Our secondary endpoint was the proportion with at least one mammogram within 24 months of surgery, performed in a subset who maintained their insurance coverage for at least that duration. Results: The 13-month analysis included 730 men with a median age at diagnosis of 62 years (Range: 25 to 87 years) and a median follow-up duration of 35 months (Range: 13 to 134 months). 209 (29%) of these men underwent mammography within 13 months after surgery. The characteristics of patients who underwent mammography and those who did not are shown in Table 1. Mammography was more likely after lumpectomy than mastectomy (41% vs. 27%) and after radiation therapy (41% vs. 32% in those who did not receive radiation). In a multivariate logistic regression model, more recent diagnosis (2015+) was associated with lower odds of undergoing mammography, while receipt of radiation was associated with higher odds of undergoing mammography. In the subset of patients with two or more years of post-surgery coverage (n=527), the proportion who had at least one mammogram during that 24-month period was 49% after lumpectomy and 40% after mastectomy. Conclusions: In this insured cohort, 73% of men did not undergo mammography within a year after mastectomy, and 59% did not within a year after lumpectomy. Mammography was less likely in patients diagnosed more recently (perhaps due to acknowledgment of the unique aspects of male breast cancer including a relatively low risk of contralateral second primary tumors), and more likely in those who received radiation. These variations in practice likely result from the paucity of evidence-based guidelines for male breast cancer survivorship care. More research is needed pertaining to whether or not mammograms improve clinical outcomes after curative intent treatment for male breast cancer. Table 1: Patient characteristics associated with receipt of mammogram within first 13 months after male breast cancer surgeryUnivariate AnalysisMultivariate AnalysisNo Mammogram (N=521)Mammogram (N=209)P-value, chi-square testOdds Ratio (OR) and 95% CIP-value for ORAge Group:0.1225-4966 (12.7%)32 (15.3%)Reference50-64216 (41.5%)98 (46.9%)0.99 (0.60, 1.63)0.9665-74112 (21.5%)44 (21.1%)0.87 (0.48, 1.57)0.6575+127 (24.4%)35 (16.7%)0.57 (0.30, 1.07)0.08Census Region:0.52Midwest138 (26.5%)59 (28.2%)ReferenceNortheast97 (18.6%)47 (22.5%)1.10 (0.68, 1.78)0.69South223 (42.8%)79 (37.8%)0.74 (0.49, 1.11)0.15West63 (12.1%)24 (11.5%)0.83 (0.47, 1.47)0.51Year of diagnosis:0.072007-2010126 (24.2%)65 (31.1%)Reference2011-2014199 (38.2%)82 (39.2%)0.80 (0.54, 1.20)0.292015+196 (37.6%)62 (29.7%)0.63 (0.41, 0.96)0.03Elixhauser Category:0.250148 (28.4%)69 (33.0%)Reference1-2218 (41.8%)74 (35.4%)0.85 (0.56, 1.28)0.433+155 (29.8%)66 (31.6%)1.18 (0.75, 1.87)0.47Surgery Type:0.005Lumpectomy55 (10.6%)38 (18.2%)ReferenceMastectomy466 (89.4%)171 (81.8%)1.57 (0.97, 2.55)0.07Chemotherapy:0.79No301 (57.8%)123 (58.9%)ReferenceYes220 (42.2%)86 (41.1%)0.79 (0.54, 1.16)0.23Radiation:0.02No355 (68.1%)124 (59.3%)ReferenceYes166 (31.9%)85 (40.7%)1.51 (1.03, 2.20)0.03 Citation Format: Siddhartha Yadav, Lindsey Sangaralingham, Stephanie R. Payne, Karthik V. Giridhar, Tina J. Hieken, Judy C. Boughey, Robert W. Mutter, John R. Hawse, Rafael E. Jimenez, Rachel A. Freedman, Sadia Choudhery, Fergus J. Couch, Celine M. Vachon, Nilay Shah, Roberto A. Leon-Ferre, Kathryn J. Ruddy. Surveillance mammography after treatment for male breast cancer [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P4-17-08.