Wang, Tiffany

Abstract Introduction Delirium occurrence is common and preventive strategies are resource intensive. Screening tools can prioritize patients at risk. Using machine learning, we can capture time and treatment effects that pose a challenge to delirium prediction. We aim to develop a delirium prediction model that can be used as a screening tool. Methods From the eICU Collaborative Research Database (eICU-CRD) and the Medical Information Mart for Intensive Care version III (MIMIC-III) database, patients with one or more Confusion Assessment Method-Intensive Care Unit (CAM-ICU) values and intensive care unit (ICU) length of stay greater than 24 h were included in our study. We validated our model using 21 quantitative clinical parameters and assessed performance across a range of observation and prediction windows, using different thresholds and applied interpretation techniques. We evaluate our models based on stratified repeated cross-validation using 3 algorithms, namely Logistic Regression, Random Forest, and Bidirectional Long Short-Term Memory (BiLSTM). BiLSTM represents an evolution from recurrent neural network-based Long Short-Term Memory, and with a backward input, preserves information from both past and future. Model performance is measured using Area Under Receiver Operating Characteristic, Area Under Precision Recall Curve, Recall, Precision (Positive Predictive Value), and Negative Predictive Value metrics. Results We evaluated our results on 16 546 patients (47% female) and 6294 patients (44% female) from eICU-CRD and MIMIC-III databases, respectively. Performance was best in BiLSTM models where, precision and recall changed from 37.52% (95% confidence interval [CI], 36.00%–39.05%) to 17.45 (95% CI, 15.83%–19.08%) and 86.1% (95% CI, 82.49%–89.71%) to 75.58% (95% CI, 68.33%–82.83%), respectively as prediction window increased from 12 to 96 h. After optimizing for higher recall, precision and recall changed from 26.96% (95% CI, 24.99%–28.94%) to 11.34% (95% CI, 10.71%–11.98%) and 93.73% (95% CI, 93.1%–94.37%) to 92.57% (95% CI, 88.19%–96.95%), respectively. Comparable results were obtained in the MIMIC-III cohort. Conclusions Our model performed comparably to contemporary models using fewer variables. Using techniques like sliding windows, modification of threshold to augment recall and feature ranking for interpretability, we addressed shortcomings of current models.

Abstract BACKGROUND: nal-IRI is a liposomal formulation of irinotecan (topoisomerase-1 inhibitor). Preclinical data show that nal-IRI accumulates in BMs and prolongs survival in animal models of BM. Findings from a phase 1 expansion study (NCT01770353), evaluating patients with mBC and active BM, are reported. METHODS: This phase 1 expansion study enrolled patients with mBC who received multiple prior lines of cytotoxic therapy in the metastatic setting, including one cohort with mBC and active BM, defined as radiographic evidence of new or progressive central nervous system (CNS) metastases after radiation therapy with ≥1 lesion of ≥1 cm in the longest dimension on gadolinium-enhanced magnetic resonance imaging. Patients received nal-IRI 50 mg/m2 (free-base equivalent; FBE) every two weeks (q2w) as an intravenous infusion over 90 minutes, escalating to 70 mg/m2 FBE q2w, if tolerated. RECIST v1.1 and modified RECIST criteria were used to assses non-CNS and CNS disease, respectively. RESULTS: In total, 30 patients were enrolled (10 with active BM). Median age was 53 years (range 29–70 years) and median number of prior cytotoxic anti-cancer regimens was 3 (range 0–6); 29 patients received ≥1 dose of nal-IRI 50 mg/m2 FBE. Overall, nal-IRI monotherapy appeared to be well tolerated, and achieved ≥30% objective response rates for both CNS and non-CNS disease. Among the 10 patients with active BM, 6 achieved CNS disease control (3 partial responses [PRs] and 3 stable disease [SD]), including one patient with durable CNS SD and non-CNS PR for 2 years. Among 7 patients with serial evaluation of CNS metastases posttreatment, 6 patients achieved a reduction in target CNS lesions compared with baseline. CONCLUSION: Treatment with nal-IRI resulted in CNS disease control among 6 of 10 heavily pretreated patients with mBC and active BM. Further exploration of nal-IRI in patients with mBC and active BM is warranted.

Abstract Background nal-IRI is a liposomal formulation of irinotecan (topoisomerase-1 inhibitor), using intraliposomal stabilisation technology to enable high drug load and in-vivo stability. The expansion of Phase I NCT01770353 evaluated nal-IRI in patients (pts) with mBC. Methods This Phase I expansion study enrolled mBC pts (Cohort 1: ER+ and/or PR+ HER2 BC (C-1); Cohort 2: Triple Negative BC (TNBC, C-2); Cohort 3: BC with active Brain Metastasis (C-3)). Key inclusion criteria: ECOG ≤1; adequate organ function; received >1 to ≤5 prior lines of cytotoxic therapy in metastatic setting, except TNBC pts who could have progressed within 12 months of adjuvant therapy. Pts received nal-IRI 50 mg/m2 (Free-base equivalent, FBE) q2w iv infusion, escalating to 70 mg/m2 FBE q2w, if tolerated. Results 30 pts (10 per cohort) were enrolled: 29 received >1 dose nal-IRI 50 mg/m2 FBE (median age 53 yrs, range 29-70 yrs). Median number of cytotoxic therapies prior to study entry was 3. The most frequent treatment emergent adverse events (TEAEs) were gastrointestinal disorders (all cohorts, 100%), primarily diarrhea (C-1, 100% / C-2, 100% / C-3, 70%), nausea (C-1, 80% / C-2, 56% / C-3, 30%), and vomiting (C-1, 60% / C-2, 33% / C-3, 30%). The incidence of neutropenia was low (C-1, 10% / C-2, 0% / C-3, 10%), all grade 1 / 2. Most common grade 3 TEAEs were diarrhea (28%), nausea (17%), fatigue (14%), hypokalemia (10%) and asthenia (10%), with none reported as grade 4. No grade 5 TEAEs were reported. Partial response (per RECIST) was observed in 10 pts for an objective response rate (ORR) of 35% (C-1, 40% / C-2, 33%) / C-3, 30%). Stable disease was observed in 5 pts (17%). Duration of Response (median, months) was C-1, 7.5 (6.4-13.0) / C-2, 5.6 (3.7-7.4) / C-3, 4.1 (0.0-22.2, extracranial: RECIST) / C-3, 1.8 (0.0-1.9, intracranial: mRECIST). Table 1.Study Demographics, Adverse Events, and Best Overall ResponseDemographicsCohort 1ER+ and/or PR+ Breast Cancern=10Cohort 2Triple Negative Breast Cancern=10Cohort 3Breast Cancer with Active Brain Metastasisn=10Total PopulationN=30Gender, Female, n (%)10 (100%)10 (100%)10 (100%)30 (100%)Age, Years Median (range)56 (49-68)52.5 (37-70)45.5 (29-63)53 (29-70)Race, n (%)WhiteBlack/AfricanOther8 (80%)02 (20%)8 (80%)1 (10%)1 (10%)7 (70%)1 (10%)2 (20%)23(77%)2 (7%)5(16%)Months Since Metastatic diagnosis, Months Median (Range)63.7 (16-87)20.7 (0-34)32.4 (8-55)24.0 (0-87)Number of Prior Cytotoxic Anti-cancer Regimens Median (Range)3.0 (1-6)3.0 (0-5)32.4 (1-6)3.0 (0-6)Safety PopulationCohort 1ER+ and/or PR+ Breast Cancern=10Cohort 2Triple Negative Breast Cancern=9Cohort 3Breast Cancer with Active Brain Metastasisn=10Total PopulationN=29ExposureTreatment Duration, Weeks Median (range)6.1 (0-78)12.3 (0-49)13.9 (1-105)12.3 (0-105)Adverse EventsAny TEAE10 (100%)9 (100%)10 (100%)29 (100%)Grade 38 (80%)6 (67%)10 (100%)20 (69%)Grade 41 (10%)06 (60%)3 (10%)TEAE Related to nal-IRI10 (100%)8 (89%)10 (100%)28 (97%)Grade > 36 (60%)3 (33%)3 (30%)12 (41%)TEAE Related to nal-IRILeading to Dose Adjustment*7 (70%)3 (33%)4 (40%)14 (48%)SAEs6 (60%)4 (44%)7 (70%)17 (59%)SAE Related to nal-IRI4 (40%)1 (11%)1 (10%)6 (21%)Duration of Objective ResponseRECIST 1.1RECIST 1.1RECIST 1.1mRECISTMonths Median (Range)7.5 (6.4-13.0)5.6 (3.7-7.4)4.1 (0.0-22.2)1.84 (0.0-1.9)Best Overall ResponseRECISTRECISTRECISTmRECISTComplete Response (CR)0000Partial Response (PR)4 (40%)3 (33%)3 (30%)3 (30%)Stable Disease (SD)03 (33%)2 (20%)3 (30%)Progressive Disease (PD)5 (50%)2 (22%)3 (30%)2 (20%)Not Evaluable (NE)1 (10%)1 (11%)2 (20%)2 (20%)Objective Response RateCR+PR(95% CI)4 (40%)12.2 -73.83 (33%)7.5-70.73 (30%)6.7-65.33 (30%)(6.7-65.3)*n=1 patient in Cohort 1 discontinued treatment due to TEAE related to treatment with nal-IRI CI, Confidence Interval; nal-IRI, Irinotecan Liposome Injection; SAE, Serious Adverse Event; TEAE, Treatment Emergent Adverse EventSystemic tumor response was evaluated according to RECIST version 1.1CNS tumor response was evaluated according to mRECIST (Anders C. et al., Breast Cancer Res Treat, 2014: 146(3):557-66) Conclusion nal-IRI monotherapy 50g/m2 FBE Q2W appears well tolerated, and achieved >30% ORR for systemic and CNS disease among heavily pre-treated mBC pts regardless of receptor status. Citation Format: Jasgit Sachdev, Pamela Munster, Donald Northfelt, Hyo S. Han, Cynthia MA, Fiona Maxwell, Tiffany Wang, Bruce Belanger, Bin Zhang, Yan Moore, Carey Anders. Phase I expansion study of irinotecan liposome injection (nal-IRI) in patients with metastatic breast cancer (mBC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr CT048.