Clinical and Genomic Characterization of Long-Term Responders Receiving Immune Checkpoint Blockade for Metastatic Non-Small-Cell Lung Cancer

OBJECTIVES

Understand from a real-world cohort the unique clinical and genomic determinants of a durable response to immune checkpoint inhibitors (ICIs).

MATERIALS AND METHODS

This is a retrospective study of patients with NSCLC who received any ICI-based regimen as first or second line therapy. Long-term responders (LTR) achieved an overall survival (OS) ≥ 3 years from time of treatment start, while nonresponders (NR) were patients who had an OS of 6 to 12 months from time of treatment start. Clinical and demographic covariables were collected from electronic medical records. Fisher's exact test and Mann-Whitney test were used to analyze the association of a long-term response to ICI in relation to clinical and genomic variables. All P-values were considered significant at P-value < .05.

RESULTS

A total of 72 patients were included in this study (LTR n = 37, NR n = 35). There were no significant differences in age, sex, race, and BMI between groups. The presence of liver metastases at the time of ICI initiation and PD-L1 status were not associated with LTR to ICIs. Patients in the LTR were more likely to experience irAEs at 3-,6- and 12-months. KRAS mutant tumors were numerically more common in the LTR group (n = 13 vs. 8).

CONCLUSION

We observe no strong clinical and biomarkers of a prolonged response to ICIs. Additional large prospective cohort studies are needed to investigate the genomic footprint of long-term responders.

Lung Cancer Survivorship: Physical, Social, Emotional, and Medical Needs of NSCLC Survivors

BACKGROUND

Newer therapies prolong survival for patients with lung cancer. Beyond extending survival, the needs of lung cancer (LC) survivors are poorly described.

METHODS

We conducted a single-institution needs assessment survey of LC survivors alive ≥1 year from diagnosis. Needs were rated on a 5-point Likert scale for 4 domains (physical, social, emotional, and medical). Multiple regression models identified demographic or treatment characteristics associated with more needs in each category. A subset analysis of survivors with metastatic LC was performed.

RESULTS

Of 360 patients approached, 235 surveys were completed. Among completed survey respondents, the median age was 69 years; most were female (62%), married (71%), and White (74%); and 41% had stage IV cancer. Finding support resources (34%) was the most common medical need. Fatigue (70%), sleep disturbance (60%), memory and concentration (57.5%), weakness (54%), and trouble breathing (51%) were physical needs affecting more than half of respondents. The most common social need was managing daily activities (42%). Emotional needs were highly prevalent, with 79% of respondents reporting a fear of recurrence and 74.5% reporting living with uncertainty. Multiple regression analysis identified that receipt of multiple lines of systemic therapy and lower household income were associated with higher physical and social needs. Younger age was associated with having a greater number of social and emotional needs. Similar results were found in the subset of survivors with metastatic disease at diagnosis.

CONCLUSIONS

The needs of LC survivors are diverse across multiple domains. Several clinical and demographic factors are independently associated with higher numbers of patient-reported needs. Our study identifies critical gaps in survivorship care for LC survivors with all stages of disease and highlights areas of future intervention.

Infectious Complications in Patients With Non-small Cell Lung Cancer Treated With Immune Checkpoint Inhibitors

INTRODUCTION

Immune checkpoint inhibitors (ICI) are standard treatment for nonsmall cell lung cancer (NSCLC). However, the burden of infectious complications during ICI therapy is poorly described.

MATERIALS AND METHODS

We conducted a retrospective study of patients with NSCLC treated with ICIs between 2007 and 2020 at a tertiary academic center. The incidence, characteristics, and healthcare utilization outcomes of infections during ICI therapy and within 3 months of ICI discontinuation are presented using descriptive statistics. Cox proportional hazard models are used to examine infection-free survival by demographic and treatment factors. Associations between patient or treatment characteristics and hospitalization or ICU admission are analyzed by logistic regression, presented as odds ratios (OR).

RESULTS

Of 298 patients, infections occurred in 54.4% (n = 162). Of these patients, 59.3% (n = 96) required hospitalization and 15.4% (n = 25) required ICU admission. The most common infection was bacterial pneumonia. Fungal infections occurred in 12 patients (7.4%). Patients with chronic obstructive pulmonary disease (COPD) (OR 2.15, 95% CI, 1.01-4.58), corticosteroid treatment within 1 month prior to infection onset (OR 3.04, 95% CI, 1.47-6.30), and concomitant irAE and infection (OR 5.48, 95% CI, 2.15-14.00) had higher odds of hospitalization. Corticosteroid use was associated with higher odds of ICU admission (OR 3.09, 95% CI, 1.29-7.38).

CONCLUSION

In this large single-institution study we identify that more than half of patients with ICI-treated NSCLC develop infectious complications. We identify that patients with COPD, recent corticosteroid use, and concomitant irAE and infection have higher odds of hospitalization, and that unusual infections (eg, fungal) can occur. This highlights clinical awareness of infections as important complications during ICI therapy in patients with NSCLC.

Cell-free DNA measurable residual disease as a predictor of postallogeneic hematopoietic cell transplant outcomes

The measurable residual disease (MRD) assessment provides an attractive predictor of allogeneic hematopoietic cell transplnat (alloHCT) outcomes. Cell-free DNA (cfDNA) has been applied to diagnosis, early detection, and disease burden monitoring in various tumors, but its utility as an MRD test in myeloid malignancies has not been systematically evaluated. We sought to determine the differential sensitivity between bone marrow (BM) and cfDNA MRD and to assess the effect of cfDNA MRD on alloHCT outcomes. The technical and clinical validation cohorts, including 82 patients participating in clinical trials (Bone Marrow Transplant Clinical Trials Network-0201 and 0402), were used. Ultradeep error-corrected targeted sequencing was performed on plasma and BM-derived DNA. We demonstrated that 94.6% (range, 93.9-95.3) of cfDNA was derived from hematopoietic tissue. The mutant allele fraction was congruent between BM and cfDNA (rho = 0.8; P < .0001); however, cfDNA seemed to be more sensitive in detecting clones with a variant allele frequency (VAF) of <0.26%. cfDNA-MRD clearance by day 90 after alloHCT (D90) was associated with improved relapse-free survival (RFS, median survival not reached vs 5.5 months; P < .0001) and overall survival (OS, median survival not reached vs 7.3 months; P < .0001) when compared with patients with persistent MRD. Irrespective of pre-alloHCT MRD, D90 cfDNA MRD was associated with inferior 2-year OS (16.7% vs 84.8%; P < .0001) and RFS (16.7% vs 80.7%; P < .0001). cfDNA seems to be an accurate, minimally invasive alternative to BM aspirates in MRD assessment and confers important prognostic implications in patients with myeloid malignancies undergoing alloHCT.

Adagrasib: a novel inhibitor for KRASG12C-mutated non-small-cell lung cancer

Adagrasib is a recently US FDA-approved novel KRAS^G12C targeted therapy with clinical efficacy in patients with advanced, pretreated KRAS^G12C-mutated non-small-cell lung cancer. KRYSTAL-I reported an objective response rate of 42.9% with median duration of response of 8.5 months. Treatment-related adverse events were primarily gastrointestinal and occurred in 97.4% of patients, with grade 3+ treatment-related adverse events occurring in 44.8% of patients. This review details the preclinical and clinical data for adagrasib in the treatment of non-small-cell lung cancer. We also outline practical clinical administration guidelines for this novel therapy, including management of toxicities. Finally, we discuss the implications of resistance mechanisms, summarize other KRAS^G12C inhibitors currently in development and outline future directions for adagrasib-based combination therapies.

Two Cases of Subsequent Hepatocellular Carcinoma in Immune Checkpoint Inhibitor-Responsive NSCLC: A Case Report

As novel therapeutic regimens continue to lead to increased survival of patients with lung cancer, it is imperative to remain mindful of the accompanying increase in the incidence of new primary malignancies. Although the most common secondary malignancies in patients with lung cancer have historically included colon, rectal, esophageal, and thyroid cancers, we report here two rare cases of new primary hepatocellular carcinomas in patients receiving immune checkpoint inhibitor therapy for NSCLC. In both cases, the diagnosis of hepatocellular carcinoma, rather than assuming a hepatic metastasis, was crucial for determining the appropriate approach for treatment. These cases thus underscore the importance of appropriate diagnostics to ensure that the proper therapeutics are chosen and present important considerations for the lung cancer community going forward.

Definitive Chemoradiation and Durvalumab Consolidation for Locally Advanced, Unresectable KRAS-mutated Non-Small Cell Lung Cancer

BACKGROUND

Consolidation durvalumab immunotherapy following definitive chemoradiation (CRT) for unresectable stage III non-small cell lung cancer (NSCLC) improves overall survival. As therapeutic options for patients with KRAS-driven disease evolve, more understanding regarding genomic determinants of response and patterns of progression for durvalumab consolidation is needed to optimize outcomes.

METHODS

We conducted a single-institutional retrospective analysis of real-world patients with locally advanced, unresectable NSCLC who completed CRT and received durvalumab consolidation. Kaplan-Meier analyses compared progression-free survival (PFS) and overall survival (OS) from start of durvalumab consolidation between patients with KRAS-mutated and non-mutated tumors. Fisher's exact test was used to compare rates of intrathoracic or extrathoracic progression.

RESULTS

Of 74 response-evaluable patients, 39 had clinical genomic profiling performed. 18 patients had tumors with KRAS mutations, 7 patients had tumors with non-KRAS actionable alterations (EGFR, ALK, ERBB2, BRAF, MET, RET, or ROS1), and 14 patients had tumors without actionable alterations. Median PFS for the overall cohort was 16.1 months. PFS for patients with KRAS-mutated NSCLC was 12.6 months versus 12.7 months for patients with non-actionable tumors (P= 0.77, log-rank). Fisher's exact test revealed a statistically significantly higher rate of extrathoracic progression versus intrathoracic-only progression for patients with KRAS-driven disease compared to patients with non-actionable tumors (P= 0.015).

CONCLUSION

Patients with KRAS-mutated NSCLC derived similar benefit from durvalumab as patients with non-actionable tumors. A higher rate of extrathoracic progression was also observed among the patients with KRAS-mutated NSCLC compared to patients with non-actionable tumors. This highlights the potential unmet needs for novel systemic therapies and surveillance methods for KRAS-mutated stage III NSCLC.

[Relationship between vitamin D receptor gene polymorphisms and gestational diabetes mellitus: a case-control study]

Objective: To investigate the relationship between vitamin D receptor (VDR) gene polymorphisms and gestational diabetes mellitus (GDM) and provide evidence for the study of the mechanism of GDM. Methods: A case-control study design was used to study pregnant women who delivered in the obstetrics department of the First Hospital of Shanxi Medical University from March 1, 2012 to July 30, 2014. Of these, 334 cases were diagnosed with GDM and were matched 1∶1 by age, gestation time and residence to corresponding healthy controls. DNA genotyping was performed for the study subjects, and those with genotyping deletions >10% were excluded. Finally 323 cases and 320 controls were included in the study. Under co-dominant, dominant, recessive, and allele genetic models, unconditional logistic regression analysis on the relationship between VDR gene locus polymorphism and GDM was conducted. And software Haploview was used to analyze the relationship between haplotype and GDM. Results: At the genetic level, VDR gene was associated with the risk of developing GDM (P<0.05). After adjusting for pre-pregnancy body mass index, family history of diabetes, it was found that rs7967152 loci was associated with an increased risk of developing GDM (AC vs. AA, OR=1.58, 95%CI: 1.13-2.21; AC+CC vs. AA, OR=1.58, 95%CI: 1.15-2.18; C vs. A, OR=1.41, 95%CI: 1.10-1.82) and rs2238140 loci was associated with an increased risk of developing GDM (AA vs. GG, OR=2.24, 95%CI: 1.19-4.20; GA+AA vs. GG, OR=1.48, 95%CI: 1.07-2.03; A vs. G, OR=1.43, 95%CI: 1.11-1.83). Carrying rs2853564 locus AG genotype and AG+GG genotype (OR=1.46, 95%CI: 1.04-2.05; OR=1.45, 95%CI: 1.05-2.00) compared with carrying AA genotype and carrying rs2853566 locus AG genotype and AG+GG genotype (OR=1.43, 95%CI: 1.03-2.00; OR=1.41, 95%CI: 1.02-1.94) compared with carrying AA genotype were risk factors for GDM. Haplotype block consisting of rs1544410, rs7967152 in the VDR gene with GC haplotype was a risk factor for GDM(OR=1.50, 95%CI: 1.15-1.97). Conclusions: VDR gene rs7967152, rs2238140, rs2853564, rs2853566 locus polymorphisms and block (rs1544410, rs7967152) GC haplotype were associated with an incrased risk of developing GDM.

Reimagining Undergraduate Medical Education in a Post-COVID-19 Landscape

Online education due to the COVID-19 pandemic caused many medical schools to increasingly employ asynchronous and virtual learning that favored student independence and flexibility. At the same time, the COVID-19 pandemic highlighted existing shortcomings of the healthcare field in providing for marginalized and underserved communities. This perspective piece details the authors' opinions as medical students and medical educators on how to leverage the aspects of pandemic medical education to train physicians who can better address these needs.

Multiple 9-1-1 complexes promote homolog synapsis, DSB repair, and ATR signaling during mammalian meiosis

DNA damage response mechanisms have meiotic roles that ensure successful gamete formation. While completion of meiotic double-strand break (DSB) repair requires the canonical RAD9A-RAD1-HUS1 (9A-1-1) complex, mammalian meiocytes also express RAD9A and HUS1 paralogs, RAD9B and HUS1B, predicted to form alternative 9-1-1 complexes. The RAD1 subunit is shared by all predicted 9-1-1 complexes and localizes to meiotic chromosomes even in the absence of HUS1 and RAD9A. Here, we report that testis-specific disruption of RAD1 in mice resulted in impaired DSB repair, germ cell depletion, and infertility. Unlike Hus1 or Rad9a disruption, Rad1 loss in meiocytes also caused severe defects in homolog synapsis, impaired phosphorylation of ATR targets such as H2AX, CHK1, and HORMAD2, and compromised meiotic sex chromosome inactivation. Together, these results establish critical roles for both canonical and alternative 9-1-1 complexes in meiotic ATR activation and successful prophase I completion.

Targeted Treatment of Non-Small Cell Lung Cancer: Focus on Capmatinib with Companion Diagnostics

MET dysregulation promoting tumorigenesis in non-small cell lung cancer (NSCLC) is associated with worse outcomes following chemotherapy as compared to non-driver mutated NSCLC and occurs either through mutations causing MET exon 14 skipping (METex14) or gene amplification and overexpression that result in enhanced receptor signaling. Capmatinib is the first FDA-approved targeted therapy for NSCLC with METex14 skipping mutations, approved in 2020. FoundationOne^® CDx, a comprehensive genomic profiling test for solid tumors, was concurrently approved as a companion diagnostic for capmatinib use. The GEOMETRY mono-1 phase II trial of capmatinib monotherapy demonstrated an overall response rate (ORR) of 68% in treatment naïve (n=28) and 41% in pre-treated (n=69) METex14 skipping advanced NSCLC; in MET amplified advanced NSCLC (gene copy number ≥ 10) ORRs of 40% in treatment naïve and 29% in pre-treated disease was seen. This review outlines the clinical data supporting capmatinib approval in the treatment of NSCLC and FoundationOne^® CDx approval as a companion diagnostic. We detail the practical clinical administration of capmatinib, including dosing and toxicity management, compare capmatinib to other approved and investigational MET-targeted therapies, discuss limitations of capmatinib, and highlight ongoing trials of capmatinib in combinatorial approaches.

[Relationship between the pre-pregnancy BMI, gestational weight gain, and risk of preeclampsia and its subtypes]

Objective: To explore the effects of maternal pre-pregnancy body mass index (BMI) and gestational weight gain and its subtypes on the risk of preeclampsia. Methods: Pregnant women delivered in the Department of Obstetrics and Gynecology of the First Affiliated Hospital of Shanxi Medical University from March 2012 to September 2016 were selected as the research subjects. According to the inclusion and exclusion criteria, 9 274 pregnant women were included. 901 preeclampsia pregnant women were selected as the case group, and 8 373 non-preeclampsia pregnant women were selected as the control group. General demographic characteristics, pre-pregnancy weight, height, lifestyle during pregnancy, reproductive history, and disease history of pregnant women were collected, and pre-pregnancy BMI and gestational weight gain were calculated. Unconditional logistic regression was used to analyze the relationship between pre-pregnancy BMI and weight gain during pregnancy and PE and its clinical subtypes. Results: Among the 901 preeclampsia after inclusion and exclusion, 401 cases were diagnosed as early-onset PE (EOPE), 500 cases were late-onset PE (LOPE), 178 cases were Mild PE (MPE), and 723 cases were severe PE (SPE). There were statistically significant differences between PE and non-PE pregnant women in terms of maternal age, residence, parity, family history of gestational diabetes and hypertension (P<0.05). After adjusting for the above factors, the logistic regression analysis results showed that pre-pregnancy BMI<18.5 kg/m² and inadequate gestational weight gain were protective factors for PE (OR=0.74, 95%CI: 0.56-0.98; OR=0.78, 95%CI: 0.62-0.99), while pre-pregnancy BMI≥24.0 kg/m² and excessive gestational weight gain were risk factors for PE (OR=1.82, 95%CI: 1.54-2.14; OR=1.82, 95%CI: 1.54-2.15). After subtype analysis on PE, the results showed that pre-pregnancy BMI<18.5 kg/m² was a protective factor for EOPE and MPE (OR=0.52, 95%CI: 0.32-0.83; OR=0.47, 95%CI: 0.23-0.97), while pre-pregnancy BMI≥24.0 kg/m² and excessive gestational weight gain were risk factors for clinical subtypes of PE. After stratification according to pre-pregnancy BMI, excessive gestational weight gain was the risk factor for PE (OR=1.86, 95%CI: 1.51-2.30; OR=1.90, 95%CI: 1.39-2.60) in pregnant women 18.5 kg/m²≤BMI<24.0 kg/m² and ≥24.0 kg/m². Inadequate gestational weight gain (OR=0.55, 95%CI: 0.34-0.89) was a protective factor for PE in pregnant women with pre-pregnancy BMI≥24.0 kg/m². Excessive gestational weight gain (OR=4.05, 95%CI: 1.20-13.69) was a risk factor for EOPE in pregnant women with pre-pregnancy BMI<18.5 kg/m². Excessive gestational weight gain was a risk factor for the clinical subtype of PE in pregnant women 18.5 kg/m²≤BMI<24.0 kg/m² before pregnancy. Inadequate gestational weight gain was a protective factor for EOPE and MPE (OR=0.39, 95%CI: 0.19-0.80; OR=0.29, 95%CI: 0.11-0.77) in pregnant women with pre-pregnancy BMI≥24.0 kg/m². Excessive weight gain was a risk factor for EOPE, LOPE and SPE (OR=1.60, 95%CI: 1.06-2.42;OR=2.20, 95%CI: 1.44-3.37;OR=2.28, 95%CI: 1.58-3.29). Conclusions: Pre-pregnancy BMI and gestational weight gain affect the risk of preeclampsia and its clinical subtypes. In contrast, the influence of gestational weight gain on preeclampsia varies among different pre-pregnancy BMI groups. Therefore, it is recommended to pay attention to the changes in pre-pregnancy BMI and gestational weight gain simultaneously to reduce preeclampsia.

[The regulation of tight junction protein via PKCα/β for abnormal permeability of brain microvascular endothelial cells exposed to paraquat]

Objective: To explore if conventional protein kinase C (cPKC: PKCα and PKCβ) contributes to paraquat (PQ) -induced abnormal permeability of mouse brain microvascular endothelial cells (BMECs) via the regulation of tight junction (TJ) proteins. Methods: The immortalized mouse brain endothelial cell line (bEnd.3) was used to establish a monolayer blood-brain barrier (BBB) model. In order to evaluate the function of the in vitro BBB model, the transendothelial electrical resistance (TEER) and permeability were measured by a Millicell-ERS volt-ohmmeter and sodium fluorescent (Na-FLU) , respectively. MTT assay was used to determine the relative survival rate of cells. The dose-response relationship was determined by treating cells with 0, 50, 100, 200, and 300 μmol/L PQ for 24 hours. The time-response relationship was determined by treating cells with 200 μmol/L PQ for 6, 12, 24, 48, and 72 hours. After the treatment of cells with 0, 100, 200, and 300 μmol/L PQ for 24 hours, the protein and mRNA expression levels of ZO-1, Occludin, and Claudin-5 were measured by immunofluorescence (IF) and quantitative real-time RT-PCR (qRT-PCR) , respectively; the expression of PKCα, PKCβ, phosphorylated (p) -PKCα, and p-PKCβ was determined by Western blot. After the treatment of cells with 200? mol/L PQ for 24 hours following the pretreatment with a classical PKC inhibitor (Go 6983, 1 μmol/L) for 1 hour, the protein expression of ZO-1, Occludin, Claudin-5, p-PKCα, and p-PKCβ was measured by Western blot. Results: The TEER of the bEnd. 3 cells increased gradually with the cell culture time, and reached a peak value of 114.3±6.9 Ω·cm(2) on day 6. According to the permeability analysis by Na-FLU, cell permeability gradually decreased with the cell culture time, and reached 1.7±0.2 cm/min on day 6, suggesting a well-behaved barrier function of cells. Compared with the control group, the survival rates of the bEnd.3 cells were significantly reduced after exposure to 100, 200, or 300 μmol/L PQ for 24 hours (P<0.05) , or after exposure to 200 μmol/L PQ for 6, 12, 24, 48, or 72 hours (P <0.05) , indicating a dose-and time-dependent relationship. The IF and qRT-PCR results showed that the protein and mRNA expression levels of ZO-1, Occludin, and Claudin-5 were significantly reduced with the increase in the concentration of PQ (P<0.05) . The Western blot analysis showed that compared with the control group, cells exposed to PQ had significantly higher protein expression of p-PKCα and p-PKCβ and significantly lower protein expression of ZO-1, Occludin, and Claudin-5 (P<0.05) . Compared with the PQ treatment group, the Go 6983 intervention group had significantly higher protein expression of ZO-1, Occludin, and Claudin-5 and significantly lower protein expression of p-PKCα and p-PKCβ (P<0.05) . Conclusion: By activation of cPKC (PKCα and PKCβ) , PQ reduces the protein and mRNA expression of TJ proteins and enhances the permeability of murine BMECs.

[Role of MAPK signaling pathway in epithelial-mesenchymal transition of type II alveolar epithelial cells induced by Paraquat]

Objective: To investigate the roles of p38 mitogen-activated protein kinases (p38 MAPK) , extracellular regulated protein kinases (ERK) and c-Jun N-tenninal kinases (JNK) of MAPK signaling pathway in Paraquat-induced epithelial to mesenchymal transition (EMT) of type II alveolarepithelial cells. Methods: RLE-6NT cells were incubated with different concentrations of PQ (0, 25, 50, 100μmol/L) for 6, 12 and 24 h. Cell morphology alteration was observed under phase-contrast microscopy. Cell viability was determined using an MTT assay. Cell migration ability was detected using scratch wound assay. Protein expression of P-p38 MAP, P-Erk1/2, P-JNK, E-cad, ZO-1, Vimentin and а-SMA were detected by western blot. The level of genes related to fibrosis (COL-I, COL-III, FN and FSP-1) were analyzed via quantitative real-time RT-PCR. Results: Cell morphology started to undergo EMT changes with a phenotype characteristic of mesenchymal cells, including an elongated shape and a lack of tight cell-cell adhesions induced by 100μmol/L PQ treatment in a time-dependent manner. MTT showed that cell viability decreased with increasing PQ concentration (50、100、200、300 μmol/L PQ treatment for 24 h) and increasing treatment time (200 μmol/L PQ treatment for 6, 12, 24, 36, 48 h) . Compared to control group, the expressions of the epithelial phenotype marker E-cad and ZO-1 significantly decreased with PQ treatment (50, 100μmol/L) in a time-dependent manner (P<0.05) . Additionally, the level of the mesenchymal marker (a-SMA, vimentin) dramatically increased with PQ treatment in the same concentration-and time-dependent manner (P<0.05) . Cell migration ability was markedly increased after 24 h of 100 μmol/L PQ treatment compared to control (P<0.05) . The phosphorylated forms of p38 MAPK, Erk1/2, and JNK were increased at 24 h after stimulation with PQ (P<0.05) . This PQ induced (100 μmol/L) phosphorylation was markedly attenuated in the presence of the p38 MAPK, ERK and JNK inhibitors (SB-203580, SP-600125 and PD98059) respectively. Furthermore, RT-PCR showed that PQ significantly induced the upregulation expression of COL I and III mRNA, Fn, and FSP-1 mRNA (P<0.05) . Conclusion: PQ-induced pulmonary fibrosis occurs via EMT, which is mediated by the MAPK pathway.

imDC: an ensemble learning method for imbalanced classification with miRNA data

Imbalances typically exist in bioinformatics and are also common in other areas. A drawback of traditional machine learning methods is the relatively little attention given to small sample classification. Thus, we developed imDC, which uses an ensemble learning concept in combination with weights and sample misclassification information to effectively classify imbalanced data. Our method showed better results when compared to other algorithms with UCI machine learning datasets and microRNA data.

Genetic algorithm-based efficient feature selection for classification of pre-miRNAs

In order to classify the real/pseudo human precursor microRNA (pre-miRNAs) hairpins with ab initio methods, numerous features are extracted from the primary sequence and second structure of pre-miRNAs. However, they include some redundant and useless features. It is essential to select the most representative feature subset; this contributes to improving the classification accuracy. We propose a novel feature selection method based on a genetic algorithm, according to the characteristics of human pre-miRNAs. The information gain of a feature, the feature conservation relative to stem parts of pre-miRNA, and the redundancy among features are all considered. Feature conservation was introduced for the first time. Experimental results were validated by cross-validation using datasets composed of human real/pseudo pre-miRNAs. Compared with microPred, our classifier miPredGA, achieved more reliable sensitivity and specificity. The accuracy was improved nearly 12%. The feature selection algorithm is useful for constructing more efficient classifiers for identification of real human pre-miRNAs from pseudo hairpins.

Mining SNPs from EST sequences using filters and ensemble classifiers

Abundant single nucleotide polymorphisms (SNPs) provide the most complete information for genome-wide association studies. However, due to the bottleneck of manual discovery of putative SNPs and the inaccessibility of the original sequencing reads, it is essential to develop a more efficient and accurate computational method for automated SNP detection. We propose a novel computational method to rapidly find true SNPs in public-available EST (expressed sequence tag) databases; this method is implemented as SNPDigger. EST sequences are clustered and aligned. SNP candidates are then obtained according to a measure of redundant frequency. Several new informative biological features, such as the structural neighbor profiles and the physical position of the SNP, were extracted from EST sequences, and the effectiveness of these features was demonstrated. An ensemble classifier, which employs a carefully selected feature set, was included for the imbalanced training data. The sensitivity and specificity of our method both exceeded 80% for human genetic data in the cross validation. Our method enables detection of SNPs from the user's own EST dataset and can be used on species for which there is no genome data. Our tests showed that this method can effectively guide SNP discovery in ESTs and will be useful to avoid and save the cost of biological analyses.

The mechanical and biological properties of demineralised cortical bone allografts in animals

We studied the calcium content and mechanical strength of cortical bone from rats and dogs after different periods of demineralisation, showing that the rate of demineralisation differed considerably between the species. Specimens from the rat were further treated by chemical extraction and autolysis and tested for osteoinductive properties. We showed that partially demineralised cortical bone retained adequate mechanical strength, while retaining the biological effects of completely demineralised bone. This shows that it is possible to prepare allografts which have adequate mechanical strength and still retain osteo-inductive properties.