Miller, KD

Abstract Objectives Patients referred for specialized care often arrive with outside medical records (OMRs) compiled into multi-report PDFs that include imaging, pathology, and clinical notes in unstructured formats. Reviewing these records is time consuming and mentally taxing, increasing the risk of delayed care, clinician frustration, and missed information affecting quality of care. This study aimed to automate the segmentation, classification, and date extraction of scanned OMRs, with a focus on records relevant to breast cancer care. Materials and Methods We used optical character recognition (OCR) to extract machine-readable text from 1303 scanned PDF documents from 116 distinct external institutions. Gemini 1.5, a large language model (LLM), was then used to segment multi-report files into individual documents, classify them into clinically meaningful categories such as mammograms and pathology reports, and extract study dates to build diagnostic timelines. Document categories were informed by clinical workflows in a breast cancer center. Results The system achieved an F1 score of 0.95 for segmentation, 0.96 for classification, and 0.90 for date extraction. In a pilot of 45 records reviewed by clinicians, only 2 classification errors and 1 date error were reported. Clinicians estimated that the tool reduced OMR review time by 40%, improved workflow efficiency, and increased satisfaction. Discussion Our findings demonstrate that combining OCR with LLMs can significantly enhance the processing of unstructured medical records, reducing manual burden and supporting timely clinical decision-making. Conclusion This study demonstrates the successful application of OCR and LLMs for organizing scanned OMRs within a specialty clinic. By automating a previously manual process, the approach supports scalable review of incoming outside records and has potential for adaptation to other clinical workflows. Future work will focus on evaluating the system across additional specialties and institutions.

Abstract Background LIV-1, a transmembrane protein and downstream target of STAT3, is highly expressed in breast cancer cells. It is associated with lymph node involvement and metastatic progression. SGN-LIV1A is an anti-LIV-1 antibody conjugated via a protease-cleavable linker to monomethyl auristatin E (MMAE). Upon binding to cell-surface LIV-1, SGN-LIV1A is internalized and releases MMAE, which disrupts microtubulin and induces apoptosis. Methods This ongoing, phase 1 study evaluates safety, tolerability, pharmacokinetics, and antitumor activity of SGN-LIV1A (q3wks IV) in women with LIV-1-positive, unresectable, locally advanced or metastatic breast cancer (LA/MBC) (NCT01969643). Patients (pts) with measurable disease and ≥2 prior cytotoxic regimens for LA/MBC are eligible. Pts with ≥ Grade 2 neuropathy are excluded. Response is assessed per RECIST v1.1; pts with stable disease (SD) or better can continue treatment until disease progression or intolerable toxicity. At completion of dose escalation in hormone receptor-positive/HER2-negative (HR+/HER2–) and triple-negative (TN) pts, expansion cohorts were opened to further evaluate safety and antitumor activity of monotherapy in TN pts. Tumor biopsies are evaluated for LIV-1 expression. Results To date, 69 pts (18 HR+/HER2–, 51 TN) have received a median of 3 cycles (range, 1–12) of SGN-LIV1A at doses of 0.5–2.8 mg/kg. Median age was 56 yrs. Pts had a median of 3 prior cytotoxic regimens for LA/MBC; 58 had visceral disease and 37 had bone metastases. No dose-limiting toxicities (DLTs) occurred in 19 DLT-evaluable pts; maximum tolerated dose was not exceeded at 2.8 mg/kg. Expansion cohorts of TN pts were opened at 2.0 and 2.5 mg/kg. Treatment-emergent adverse events (AEs) reported in ≥25% of pts were fatigue (59%), nausea (51%), peripheral neuropathy (44%), alopecia (36%), decreased appetite (33%), constipation (30%), abdominal pain, diarrhea, and neutropenia (25% each). Most AEs were Grade 1/2; AEs ≥ Grade 3 included neutropenia (25%) and anemia (15%). Febrile neutropenia occurred in 2 pts whose total dose exceeded 200 mg per cycle, including 1 treatment-related death due to sepsis. No other treatment-related deaths occurred on-study. Seven pts discontinued treatment due to AEs. In dose escalation, activity was observed in 17 efficacy evaluable (EE) HR+/HER2- pts, with a disease control rate (DCR= CR+PR+SD) of 59% (10 SD), including 1 pt with SD ≥24 wks. Among the 44 EE TN pts (dose escalation plus expansion cohorts), the objective response rate (ORR) was 32% (14 PR) with a confirmed PR rate of 21%, DCR was 64% (14 PR, 14 SD), and clinical benefit rate (CBR=CR+PR+SD ≥24 wks) was 36% (16 pts). For TN pts, median PFS was 11.3 wks (95% CI: 6.1, 17.1); 10 pts remain on treatment. Of 631 MBC tumor samples of all clinical subtypes evaluated for LIV-1, 91% were positive; 75% had moderate-to-high expression (H-score ≥100). Conclusions LIV-1 is expressed in almost all MBC tumors. SGN-LIV1A monotherapy was generally well tolerated and showed encouraging antitumor activity in heavily pretreated TN MBC, with a PR rate of 32%, confirmed PR rate of 21%, and CBR (≥24 wks) of 36%. Response duration data continue to evolve. Enrollment continues in the TN monotherapy expansion cohort. Citation Format: Modi S, Pusztai L, Forero A, Mita M, Miller KD, Weise A, Krop I, Burris III H, Kalinsky K, Tsai M, Liu MC, Hurvitz SA, Wilks S, Ademuyiwa F, Diab S, Han HS, Kato G, Nanda R, O'Shaughnessy J, Kostic A, Li M, Specht J. Phase 1 study of the antibody-drug conjugate SGN-LIV1A in patients with heavily pretreated triple-negative metastatic breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr PD3-14.

Abstract Background: Late recurrence 5 or more years after diagnosis accounts for least one-half of all breast cancer recurrences, especially in hormone receptor (HR)+ disease. Biomarkers prognostic for late recurrence offer potential to more accurately identify subjects who might benefit from extended adjuvant endocrine therapy, or novel strategies to reduce late recurrence risk. Methods: CTCs were assessed at a single time point using the CELL SEARCH® assay in patients without clinical evidence of recurrence between 4.5-7.5 years after an initial diagnosis of HER2- stage II-III breast cancer and enrolled in trial E5103; all patients received surgery plus adjuvant chemotherapy and endocrine therapy for at least 5 years if HR+ disease. Patients were followed for evidence of clinical recurrence in accordance with standard care, and the association between CTCs and clinical recurrence was evaluated. Results: 546 patients without clinical evidence of recurrence enrolled between 2/25/13-7/29/16 and provided a blood sample that yielded a CTC result; 16 (2.9%) subsequently had a recurrence, of whom 15 had HR-positive disease. The median time between enrollment on E5103 and CTC assay was 5.2 years, and median/mean followup after the CTC assay was 1.6 years (range 0-3.9 years). The CTC assay was positive in 27 (4.9%) (median CTC count 1/7.5 ml blood, range 1-15). There were no significant differences in patient characteristics in the CTC+ vs. CTC- cases, including age < 50 years at initial diagnosis (52% vs. 44%), tumor size > 2 cm (63% vs. 59%), >/= 1 positive axillary node (81% vs. 72%), ER and/or PR+ (70% vs. 64%) or poor histologic grade (44% vs 55%). The recurrence rate per person-year in the CTC+ vs. CTC- groups was 19.6% vs. 1.1%, respectively (P<0.01). The median/mean time to recurrence after a positive CTC assay was 2.8 years. In multivariate analysis adjusted for clinical covariates (see table), a positive CTC assay was associated with an 18.3-fold increased risk of recurrence. Univariate and Multivariate Analysis - Association Between CTC Status and RecurrenceCovariateHazard Ratio for Recurrence - Univariate Analysis (95% CI)Hazard Ratio for Recurrence - Multivariate Analysis (95% CI)Age < 50 years3.4 (0.95-11.8)3.0 (0.8-10.9)Tumor size > 2 cm4.2 (0.98-19.2)2.9 (0.6-13.2)>/= 1+ node2.6 (0.59-11.5)0.7 (0.1-3.8)ER and/or PR+8.3 (1.1-63.5)7.7 (0.7-79.9)Poor grade0.43 (0.14-1.2)1.2 (0.4-3.8)CTC -pos vs CTC-neg20.9 (7.5-58.3)18.3 (5.7-58.2) Conclusions: A single positive CTC assay in patients without clinical evidence of recurrence 5 years after diagnosis of stage II-III HR+, HER2- breast cancer provides independent prognostic information for late recurrence, providing proof of concept for using liquid-based biomarkers for late relapse risk assessment. These findings provide a foundation for further evaluation of this new risk assessment paradigm using CTC and other blood-based assays in this setting, and designing clinical trials to tailor therapeutic risk interventions. Supported by Breast Cancer Research Foundation (J. Sparano; R. Comis) and Susan G. Komen Foundation (J. Sparano), and by the National Cancer Institute (CA180820, CA180794, CA180790, CA180791, CA180795, CA180802, CA180816, CA180821, CA189859). Citation Format: Sparano JA, O'Neill A, Alpaugh K, Wolff AC, Northfelt DW, Dang C, Sledge, Jr. GW, Miller KD. Circulating tumor cells (CTCs) five years after diagnosis are prognostic for late recurrence in operable stage II-III breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr GS6-03.

Abstract Background LIV-1, a multispan transmembrane protein and downstream target of STAT3, is highly expressed in breast cancer cells (Sussman 2014). It has been associated with lymph node involvement and linked with malignant progression to metastasis (Manning 1994). SGN-LIV1A is an anti-LIV-1 antibody conjugated via a protease-cleavable linker to monomethyl auristatin E (MMAE). Upon binding to cell-surface LIV-1, SGN-LIV1A is internalized and releases MMAE, which binds to tubulin and induces G2/M arrest and apoptosis. Methods A phase 1, open-label, dose-escalation study is ongoing to evaluate the safety, tolerability, antitumor activity, pharmacokinetics (PK), and the maximum tolerated dose (MTD) of SGN-LIV1A monotherapy (q3 wks IV) in women with LIV-1-positive, metastatic breast cancer (MBC) (NCT01969643). Patients (pts) with either hormone receptor-positive/HER2-negative (HR+/HER2–) or triple-negative (TN) disease were eligible if they had measurable disease and received ≥2 prior cytotoxic regimens in the metastatic setting. Pts with ≥ Grade 2 neuropathy were excluded. Response was assessed per RECIST v1.1, and pts with stable disease (SD) or better were eligible to continue treatment until disease progression or intolerable toxicity. Fresh tumor biopsies were obtained pre- and post-treatment to evaluate the LIV-1 expression-response relationship, mechanism of action, and tumor sensitivity to SGN-LIV1A. Results To date, 21 pts (17 HR+/HER2–, 4 TN) have received a median of 3 cycles (range, 1–7) of SGN-LIV1A at doses of 0.5–2.8 mg/kg. Median age was 58 yrs (range, 34–73), and pts had a median of 8 prior systemic metastatic therapies (range, 2–15). At baseline, 19 pts had visceral disease and 15 had bone involvement. No dose-limiting toxicities (DLTs) in Cycle 1 were observed in the 18 DLT-evaluable pts; MTD has not yet been identified. Treatment-emergent adverse events (AEs) reported in ≥30% of pts were primarily Grade 1/2 in severity and were: nausea (57%), fatigue and peripheral neuropathy (43% each), alopecia (38%), and vomiting (33%). Two pts discontinued treatment due to AEs (1 nausea, 1 tachycardia). Preliminary PK data suggest a linear increase in antibody-drug conjugate (ADC) exposure with increasing dose. In the 19 efficacy-evaluable pts, the overall response rate (ORR) was 11% (2 partial responses [PRs]) with clinical benefit of SD or better achieved in 63% (2 PR, 10 SD) of pts. Of note, all 4 pts with TN MBC achieved clinical benefit: 2 PR and 2 SD. Currently, the median duration of clinical benefit is 12.7 wks (range, 6.1–26.3). Four pts remain on treatment. To date, of the 179 MBC tumor specimens evaluated for LIV-1, 156 (87%) were LIV-1-positive; moderate-to-high LIV-1 expression (H-score ≥100) was present in 91% (94 of 103) of HR+/HER2– and 81% (47 of 58) of TN samples. Conclusions LIV-1 is expressed in the majority of metastatic breast cancer tissue samples, with moderate-to-high expression in both HR+/HER2- and TN disease. To date, SGN-LIV1A monotherapy has been generally well tolerated and resulted in a clinical benefit of SD or better in 63% of these heavily pre-treated pts, including 2 PRs in the 4 TN MBC pts. Response duration data continue to evolve. Clinical subtype-specific expansion cohorts are planned. Citation Format: Forero A, Burris III H, Mita M, Specht J, Weise A, Liu MC, Modi S, Pusztai L, Kostic A, Yang J, Li M, Hengel S, Miller K. Interim analysis of a phase 1 study of the antibody-drug conjugate SGN-LIV1A in patients with metastatic breast cancer. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P3-14-05.

Abstract Background: A previous feasibility trial (E2104 – Ann Oncol 23(2):331–7,2012) suggested incorporation of B into anthracycline-containing adjuvant therapy was feasible but ongoing cardiac monitoring was required to define the true impact of B on cardiac function. Methods: Patients (pts) were assigned 1:2:2 to one of three treatment arms. In addition to doxorubicin and cyclophosphamide followed by weekly paclitaxel, patients received either placebo (Arm A – AC>T) or B during chemotherapy (Arm B - BAC>BT), or B during chemotherapy followed by B monotherapy (15 mg/kg q3wk) for an additional 10 cycles (Arm C – BAC>BT>B). Randomization was stratified and B dose adjusted for choice of AC schedule (classical q3wk − 15 mg/kg; dose dense(dd) q2 wk − 10 mg/kg). When indicated, radiation and hormonal therapy were administered concurrently with B (for Arm C pts). The primary cardiac endpoint was the incidence of clinically apparent cardiac dysfunction (CHF)defined as symptomatic decline in left ventricular ejection fraction (LVEF) to below the lower limit of normal (LLN) or symptomatic diastolic dysfunction as assessed by independent review. Cumulative toxicity data as of Jan 23, 2012 are presented. Results: From 11.07 to 2.11, 4994 pts were enrolled. Median age was 52; 80% received ddAC. Chemotherapy associated toxicities including myelosuppression (Grade 4 neutropenia 16/20/19%) and neuropathy (Grade ≥ 3 8/8/8%) were similar across all arms. Grade ≥ 3 hypertension/thrombosis/proteinuria/hemorrhage was reported by 7/3/<1/<1% of B-treated pts. 99 pts developed CHF, most commonly reported at the post-AC or post-T evaluation. After a median follow-up of 26 months, the cumulative incidence of clinical CHF at 15 months from randomization in Arm A/B/C was 1.0/1.7/2.9% respectively. Median age of CHF pts was 57; median baseline LVEF of CHF pts was 60. Conclusion: Incorporation of B into anthracycline and taxane containing adjuvant therapy results in a significant but small increase in clinical CHF. The rate of clinical CHF is similar to that predicted by E2104 (2.5–2.9%) and reported In the FDA label for anthracycline pre-treated pts(3.8%). No unexpected toxicities were encountered. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P5-17-01.