Virtual Library
Start Your Search
R. Wolfe
Author of
-
+
MA 18 - Global Tobacco Control and Epidemiology II (ID 676)
- Event: WCLC 2017
- Type: Mini Oral
- Track: Epidemiology/Primary Prevention/Tobacco Control and Cessation
- Presentations: 1
- Moderators:H. Kawai, Christian Klaus Manegold
- Coordinates: 10/17/2017, 15:45 - 17:30, Room 511 + 512
-
+
MA 18.12 - Quality of Data Informing Epidemiological Studies in Patients with Lung Cancer (ID 7550)
17:00 - 17:05 | Author(s): R. Wolfe
- Abstract
- Presentation
Background:
Epidemiological studies commonly use data from clinical (i.e. medical records) and administrative (i.e. claims data) datasets for the purposes of exploratory analyses, as well as clinical and quality reporting, benchmarking, risk adjustment, and machine learning. Validity is contingent on accurate and detailed reporting of data, demanding robust methodological validation.
Method:
Single centre retrospective comparative study assessing completeness and agreement (kappa-statistic (κ)) of data reporting for key prognostic variables across three independent data sources, among patients with lung cancer. The study population was formed by random selection of patients from an Australian single centre prospective study. Prospectively collected research study-data (SD) was extracted, and then compared to data extracted from individual patient medical records (MR) as well as International Classification of Diseases (ICD) coding from administrative data (AD).
Result:
The study population included 10% of patients from an Australian lung cancer cohort (n=111/1090), and represented the overall cohort in terms of patient demographics and disease characteristics. Prognostic data for stage, comorbidities, smoking history, performance status, and weight loss at diagnosis, was reported for >96% of patients in SD. Comparatively, AD did not report any prognostic data for 42% (47/111) of patients treated in ambulatory settings, and indeed when reported was grossly inaccurate. By way of examples, according to AD, 23% of patients had ≥1 comorbidity versus 68% by MR and 64% by SD; 38% had positive smoking history versus 78% by MR and 81% by SD; 2% had respiratory comorbidity versus 28% by MR and 37% by SD. Similar patterns were observed for other comorbid conditions. Complete TNM staging was captured in only 45% of MR at the time of first treatment, although with good concordance with SD (κ=0.9, 95%CI 0.7, 1.0). Equally when factors were documented in MR they were reasonably concordant with SD: smoking status (completeness 96.4%, κ=0.9, 95%CI 0.8, 1.0), performance status (completeness 82.0%, κ=0.5, 95%CI 0.4, 0.7) and weight loss (completeness 71.1%, κ=0.3, 95%CI 0.1, 0.5).
Conclusion:
Poor capture of factors (either omission or inaccuracy) limit the potential contribution of both MR and AD for use in clinical, epidemiological, and machine learning research – particularly when being utilised to derive diagnostic, prognostic and classification systems. Use of this data for purposes other than intended may misinform estimates of comorbidity disease burden and fail to appropriately adjust for competing mortality risks in models that inform outcomes reporting and ensuing policy decisions.
Only Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login, select "Add to Cart" and proceed to checkout. If you would like to become a member of IASLC, please click here.
-
+
P1.01 - Advanced NSCLC (ID 757)
- Event: WCLC 2017
- Type: Poster Session with Presenters Present
- Track: Advanced NSCLC
- Presentations: 1
- Moderators:
- Coordinates: 10/16/2017, 09:30 - 16:00, Exhibit Hall (Hall B + C)
-
+
P1.01-033 - Thrombogenic Biomarkers in Patients with NSCLC – Associations with Thrombosis, Progression, and Survival (ID 8852)
09:30 - 09:30 | Author(s): R. Wolfe
- Abstract
Background:
For patients with NSCLC, the risk of thromboembolism (TE) is high but heterogeneous. We aimed to profile biomarkers among NSCLC patients receiving anticancer therapy to identify patients and time periods of high TE risk where intervention with proven preventative strategies is likely to achieve maximal benefit.
Method:
We assessed the association between baseline biomarker levels and longitudinal biomarker changes, with subsequent incidence of TE (venous (VTE) or arterial (ATE)), disease progression and overall survival. Biomarkers (thromboelastography, d-dimer, fibrinogen, haemoglobin, white cell count, platelet count, neutrophil/lymphocyte ratio (NLR), and platelet/lymphocyte ratio (PLR)) were sequentially assessed at commencement of anticancer therapy (baseline), weeks 1, 4 and 12, and 3-monthly until 12 months.
Result:
During study follow-up (median 22 months, range 6-31), 129 patients were sequentially assessed over median 5 time points (range 1-9). Patients underwent surgery (n=12), chemo-radiotherapy (CRT, n=47), palliative chemotherapy (CHT, n=36), and single modality radiotherapy (RT, n=34) – only surgical patients received thromboprophylaxis. 24 patients (19%) had documented TE, 19 (15%) VTE and 5 (4%) ATE; 79% occurred within the first 6 months with median time to TE 48 days (range 1-151). Among ambulatory patients (CHT/CRT/RT), an initial model identified as high TE risk those patients with baseline fibrinogen ≥4g/L and d-dimer ≥0.5mg/L, or d-dimer ≥1.5mg/L. Hazard ratio (HR) for TE was 8.0 (p=0.04) for high versus low risk CHT/CRT patients and 6.5 (p=0.07) for high versus low risk for CHT/CRT/RT patients. Comparatively, using an established risk score, HR for TE with Khorana score ≥3 vs. <3 was 1.3 (p=0.68) for CHT/CRT and 1.1 (p=0.91) for CHT/CRT/RT. Considering temporal changes (d-dimer ≥1.5mg/L at week 4), risk assessment was enhanced with 100% sensitivity and 34% specificity for CHT/CRT. Specificity reduced to 27% when including RT patients. NLR, PLR and platelet count were not associated with TE. High TE risk patients (Fib≥4 + d-dimer ≥0.5, d-dimer ≥1.5) also had increased risk of cancer progression (HR 2.3, p<0.01) and mortality (HR 2.5, p<0.01). Baseline NLR≥2.5 and platelet count ≥350 were associated with progression (HR 2.0, p=0.01 and HR 2.0, p<0.01 respectively) and mortality (HR 2.6, p=0.01 and HR 1.9, p=0.01 respectively); PLR was not.
Conclusion:
For ambulatory patients with NSCLC, d-dimer and fibrinogen were associated with TE, cancer progression, and prognosis and could easily be applied in a simple algorithm, for real-time decision-making. In spite of low specificity, consideration of thromboprophylaxis is warranted for high risk patients given substantial TE-associated adverse clinical and economic consequences.
-
+
P3.13 - Radiology/Staging/Screening (ID 729)
- Event: WCLC 2017
- Type: Poster Session with Presenters Present
- Track: Radiology/Staging/Screening
- Presentations: 1
- Moderators:
- Coordinates: 10/18/2017, 09:30 - 16:00, Exhibit Hall (Hall B + C)
-
+
P3.13-003 - The Lung Cancer Prognostic Index – a Risk Score to Predict Overall Survival after the Diagnosis of Non-Small Cell Lung Cancer (ID 7551)
09:30 - 09:30 | Author(s): R. Wolfe
- Abstract
Background:
Outcomes in Non-Small Cell Lung Cancer (NSCLC) are poor but heterogeneous, even within TNM stage groups. To improve prognostic precision we aimed to develop and validate a simple model for the prediction of overall survival (OS) using patient and disease variables.
Method:
The study population included 1458 patients from three independent cohorts. Associations between baseline variables and OS were estimated in a derivation cohort from a prospective single-centre study (n=695) using Cox proportional hazards regression. Points were allocated to variables based on the strength of association to create the Lung Cancer Prognostic Index (LCPI). Model performance was assessed (by a c-statistic for discrimination and Cox-Snell residuals for calibration) in two independent validation cohorts (n=479 and n=284).
Result:
Three disease-related and six patient-related variables were found to predict OS: stage, histology, mutation status, performance status, weight loss, smoking history, respiratory comorbidity, sex and age. Patients were classified according to predicted LCPI score. Two-year OS rates according to LCPI in the derivation and two validation cohorts respectively were 84%, 77% and 68% (LCPI 1: score≤9); 61%, 61% and 42% (LCPI 2: score 10-13); 33%, 32% and 14% (LCPI 3: score 14-16); 7%, 16% and 5% (LCPI 4: score ≥15). Predictive performance (Harrell’s c-statistics) were 0·74 for the derivation cohort, 0·72 and 0·71 for the two validation cohorts.
Conclusion:
The LCPI contributes additional prognostic information which, in conjunction with other validated tools and evidence based management guidelines, may be applied to counsel patients, guide clinical trial eligibility, or standardise mortality risk for epidemiological analyses.
