Inferring cancer disease response from radiology reports using large language models with data augmentation and prompting.

OBJECTIVE: To assess large language models on their ability to accurately infer cancer disease response from free-text radiology reports. MATERIALS AND METHODS: We assembled 10 602 computed tomography reports from cancer patients seen at a single institution. All reports were classified into: no evidence of disease, partial response, stable disease, or progressive disease. We applied transformer models, a bidirectional long short-term memory model, a convolutional neural network model, and conventional machine learning methods to this task. Data augmentation using sentence permutation with consistency loss as well as prompt-based fine-tuning were used on the best-performing models. Models were validated on a hold-out test set and an external validation set based on Response Evaluation Criteria in Solid Tumors (RECIST) classifications. RESULTS: The best-performing model was the GatorTron transformer which achieved an accuracy of 0.8916 on the test set and 0.8919 on the RECIST validation set. Data augmentation further improved the accuracy to 0.8976. Prompt-based fine-tuning did not further improve accuracy but was able to reduce the number of training reports to 500 while still achieving good performance. DISCUSSION: These models could be used by researchers to derive progression-free survival in large datasets. It may also serve as a decision support tool by providing clinicians an automated second opinion of disease response. CONCLUSIONS: Large clinical language models demonstrate potential to infer cancer disease response from radiology reports at scale. Data augmentation techniques are useful to further improve performance. Prompt-based fine-tuning can significantly reduce the size of the training dataset.

Real world efficacy of osimertinib in second line/beyond in patients with metastatic EGFR+ non-small cell lung cancer and role of paired tumour-plasma T790M testing at tyrosine kinase inhibitor resistance.

Background: Osimertinib is a third-generation epidermal growth factor receptor tyrosine kinase inhibitor (EGFR-TKI) approved for use in EGFR-mutant lung cancer. We examined its performance in the second/subsequent line after resistance to first- and second-generation (1/2G) EGFR-TKI. Methods: We reviewed electronic records of 202 patients who received osimertinib from July 2015 to January 2019 in the second/subsequent line after progression on prior EGFR-TKI. Of these, complete data from 193 patients were available. Clinical data including patient characteristics, primary EGFR mutation, T790M mutation status, presence of baseline brain metastases (BM), first-line EGFR-TKI use, and survival outcomes were extracted, and results retrospectively analyzed. Results: Of 193 evaluable patients, 151 (78.2%) were T790M+ (T790M positive) with 96 (49.2%) tissue confirmed; 52% of patients received osimertinib in the second line setting. After median follow up of 37 months, median progression-free survival (PFS) of the entire cohort was 10.3 [95% confidence interval (CI): 8.64-11.50] months and median overall survival (OS) was 20 (95% CI: 15.61-23.13) months. Overall response rate (ORR) to osimertinib was 43% (95% CI: 35.9-50.3%); 48.3% in T790M+ vs. 20% in T790M- (T790M negative) patients. OS in T790M+ patients was 22.6 vs. 7.9 months in T790M- patients (HR 0.43, P=0.001), and PFS was 11.2 vs. 3.1 months respectively (HR 0.52, P=0.01). Tumour T790M+ was significantly associated with longer PFS (P=0.007) and OS (P=0.01) compared to tumour T790M- patients, however this association was not seen with plasma T790M+. Of the 22 patients with paired tumor/plasma T790M testing, response rate (RR) to osimertinib was 30% for those plasma T790M+/tumour T790M-, compared to 63% and 67% for those who were plasma T790M+/tumour T790M+ and plasma T790M-/tumour T790M+, respectively. By multivariable analysis (MVA), Eastern Cooperative Oncology Group (ECOG) performance status ≥2 was associated with shorter OS (HR 2.53, P<0.001) and PFS (HR 2.10, P<0.001), whereas presence of T790M+ was associated with longer OS (HR 0.50, P=0.008) and PFS (HR 0.57, P=0.027). Conclusions: This cohort demonstrated the efficacy of osimertinib in second line/beyond for EGFR+ (EGFR mutation-positive) non-small cell lung cancer (NSCLC). Tissue T790M result appeared more predictive of osimertinib efficacy compared to plasma, highlighting potential T790M heterogeneity and the advantage with paired tumor-plasma T790M testing at TKI resistance. T790M- disease at resistance remains an unmet treatment need.

Molecular testing in non-small cell lung cancer: A consensus recommendation.

Introduction: Lung cancer remains an important cause of cancer-related mortality in Singapore, with a greater proportion of non-smokers diagnosed with non-small cell lung cancer (NSCLC) in the past 2 decades. The higher prevalence of targetable genomic alterations in lung cancer diagnosed in Singapore compared with countries in the West, as well as the expanding therapeutic landscape for NSCLC in the era of precision medicine, are both factors that underscore the importance of efficient and effective molecular profiling. Method: This article provides consensus recommendations for biomarker testing for early-stage to advanced NSCLC. These recommendations are made from a multidisciplinary group of lung cancer experts in Singapore with the aim of improving patient care and long-term outcomes. Results: The recommendations address the considerations in both the advanced and early-stage settings, and take into account challenges in the implementation of biomarker testing as well as the limitations of available data. Biomarker testing for both tumour tissue and liquid biopsy are discussed. Conclusion: This consensus statement discusses the approaches and challenges of integrating molecular testing into clinical practice for patients with early- to late-stage NSCLC, and provides practical recommendations for biomarker testing for NSCLC patients in Singapore.

Dietary factors and the risk of lung cancer by epidermal growth factor receptor mutation status and histological subtypes.

Background: Previous studies have reported differential associations of certain dietary factors such as soy consumption by epidermal growth factor receptor mutant (EGFR +) subtype of non-small cell lung cancer (NSCLC). However, whether the other dietary factors including meat, fruits, and vegetables have differential risks on different histological and molecular subtypes of lung cancer remains unclear. Therefore, we conducted a case-control study to evaluate these associations. Methods: A total of 3,170 cases and 4,238 controls from three different studies (Genes and Environment in Lung Cancer Study, Lung Cancer Consortium Singapore Study, and Multi-ethnic Cohort Study) were included. Information on demographics, lifestyle, and dietary consumption was obtained using questionnaires. Diet was assessed by using the number of standard servings of each item consumed per week. Multivariable logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for the association between meat, vegetables, and fruits consumption with lung cancer risk after adjusting for potential confounders. Results: We identified a significant inverse association between higher consumption of fruits and the risk of lung cancer (2nd tertile: OR = 0.54, 95%CI = 0.46-0.65; 3rd tertile: OR = 0.77, 95%CI = 0.65-0.91), compared with the lower (1st tertile) consumption of fruits. Higher vegetable consumption was significantly associated with a lower risk of EGFR + lung cancer (OR = 0.69, 95% CI = 0.54-0.88), however, this association was not significant among EGFR wild-type (-) lung cancer. Conversely, higher consumption of total meat (OR = 2.10, 95%CI = 1.58-2.79) was significantly associated with higher lung cancer risk, as compared with the lower consumption group. Conclusions: Differential associations between vegetable consumption with EGFR mutation status in NSCLC were found. Further prospective studies are warranted to assess this association and elucidate the biological mechanisms.

Late-onset neutropenia following RCHOP chemotherapy in diffuse large B-cell lymphoma.

Rituximab has been associated with the development of late-onset neutropenia (LON). As only heterogeneous studies have been conducted, its incidence and clinical course remain unclear. We aim to: (1) study the incidence and clinical relevance of WHO grade 3/4 LON in a uniform group of patients with diffuse large B-cell lymphoma (DLBCL) in complete remission following curative rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisolone (RCHOP) chemotherapy; (2) ascertain predictive factors for LON. The 121 eligible patients identified from our prospectively maintained database were followed up for occurrence of WHO grade 3/4 LON. The clinical course of LON was documented, and its relationship with patient- and tumor-related factors was analyzed. With a median follow-up of 883 days (range, 265-1762), 13.2% had developed LON of grade 3/4. The median time to neutrophil nadir was 129 days (range, 39-277). The median time to recovery was 69 days (range, 3-349) and occurred in all except two patients. Only one episode of nonlife threatening bacterial culture-positive urinary tract infection and pulmonary tuberculosis, both occurring in the same patient was documented. Results of Fischer's exact test revealed that age, stage, LDH level, ECOG, marrow involvement, and hematologic parameters did not predict for LON development. WHO grade 3/4 LON is not infrequent in patients with DLBCL receiving RCHOP. Even so, it is reassuring that LON is self-limiting and unassociated with life-threatening infection. A watchful waiting approach is appropriate in majority of patients who develop LON following RCHOP. Am. J. Hematol., 2009. (c) 2009 Wiley-Liss, Inc.