Natural language processing of clinical notes enables early inborn error of immunity risk ascertainment.

Background: There are now approximately 450 discrete inborn errors of immunity (IEI) described; however, diagnostic rates remain suboptimal. Use of structured health record data has proven useful for patient detection but may be augmented by natural language processing (NLP). Here we present a machine learning model that can distinguish patients from controls significantly in advance of ultimate diagnosis date. Objective: We sought to create an NLP machine learning algorithm that could identify IEI patients early during the disease course and shorten the diagnostic odyssey. Methods: Our approach involved extracting a large corpus of IEI patient clinical-note text from a major referral center's electronic health record (EHR) system and a matched control corpus for comparison. We built text classifiers with simple machine learning methods and trained them on progressively longer time epochs before date of diagnosis. Results: The top performing NLP algorithm effectively distinguished cases from controls robustly 36 months before ultimate clinical diagnosis (area under precision recall curve > 0.95). Corpus analysis demonstrated that statistically enriched, IEI-relevant terms were evident 24+ months before diagnosis, validating that clinical notes can provide a signal for early prediction of IEI. Conclusion: Mining EHR notes with NLP holds promise for improving early IEI patient detection.

Convection-enhanced delivery in glioblastoma: a review of preclinical and clinical studies.

Glioblastoma is the most common malignant brain tumor, and it carries an extremely poor prognosis. Attempts to develop targeted therapies have been hindered because the blood-brain barrier prevents many drugs from reaching tumors cells. Furthermore, systemic toxicity of drugs often limits their therapeutic potential. A number of alternative methods of delivery have been developed, one of which is convection-enhanced delivery (CED), the focus of this review. The authors describe CED as a therapeutic measure and review preclinical studies and the most prominent clinical trials of CED in the treatment of glioblastoma. The utilization of this technique for the delivery of a variety of agents is covered, and its shortcomings and challenges are discussed in detail.

Quantifying Isoniazid Levels in Small Hair Samples: A Novel Method for Assessing Adherence during the Treatment of Latent and Active Tuberculosis.

BACKGROUND: Tuberculosis (TB) is the leading cause of death from an infectious pathogen worldwide and the most prevalent opportunistic infection in people living with HIV. Isoniazid preventive therapy (IPT) reduces the incidence of active TB and reduces morbidity and mortality in HIV-infected patients independently of antiretroviral therapy. However, treatment of latent or active TB is lengthy and inter-patient variability in pharmacokinetics and adherence common. Current methods of assessing adherence to TB treatment using drug levels in plasma or urine assess short-term exposure and pose logistical challenges. Drug concentrations in hair assess long-term exposure and have demonstrated pharmacodynamic relevance in HIV. METHODS: A large hair sample from a patient with active TB was obtained for assay development. Methods to pulverize hair and extract isoniazid were optimized and then the drug detected by liquid chromatography/ tandem mass spectrometry (LC/MS-MS). The method was validated for specificity, accuracy, precision, recovery, linearity and stability to establish the assay's suitability for therapeutic drug monitoring (TDM). Hair samples from patients on directly-observe isoniazid-based latent or active TB therapy from the San Francisco Department of Public Health TB clinic were then tested. RESULTS: Our LC/MS-MS-based assay detected isoniazid in quantities as low as 0.02ng/mg using 10-25 strands hair. Concentrations in spiked samples demonstrated linearity from 0.05-50ng/mg. Assay precision and accuracy for spiked quality-control samples were high, with an overall recovery rate of 79.5%. In 18 patients with latent or active TB on treatment, isoniazid was detected across a wide linear dynamic range. CONCLUSIONS: An LC-MS/MS-based assay to quantify isoniazid levels in hair with performance characteristics suitable for TDM was developed and validated. Hair concentrations of isoniazid assess long-term exposure and may be useful for monitoring adherence to latent or active TB treatment in the setting of HIV.

Incidence of headache as a presenting complaint in over 1000 patients with sellar lesions and factors predicting postoperative improvement.

INTRODUCTION: Due to the high incidence of headaches and pituitary tumors, neurosurgeons often evaluate patients with benign-appearing sellar lesions and headaches without insight into whether the headache is attributable to the lesion. We sought to evaluate the incidence of headache as a presenting complaint in patients undergoing transsphenoidal surgery for various pathologies and to identify factors predicting postoperative improvement. METHODS: We conducted a 5-year retrospective review of our first 1015 transsphenoidal surgeries since establishing a dedicated pituitary center. RESULTS: Of 1015 patients, 329 (32%) presented with headache. Of these 329 patients, 241 (73)% had headache as their chief complaint. Headache was most common in patients with apoplexy (84%), followed by Rathke's cleft cysts (RCCs) (60%). Multivariate analyses revealed diagnosis (P = 0.001), younger age (P = 0.001), and female gender (P = 0.006) to be associated with headache. Of patients presenting with headaches, 11% reported improvement at 6-week follow-up and 53% improved at 6-month follow-up. Multivariate analyses revealed gross total resection (GTR; P = 0.04) and decreased duration of headache (P = 0.04) to be associated with improvement, while diagnosis, age, gender, lesion size, whether headache was a chief complaint, and location of headache were not associated with improvement (P > 0.05). CONCLUSION: In analyzing over 1000 consecutive patients undergoing transsphenoidal surgery, younger patients, females, and patients with RCCs and apoplexy were more likely to present with headache. Patients who underwent GTR and had shorter duration of headache were more likely to experience headache improvement. This information can be used to counsel patients preoperatively.

Factors predicting recurrence after resection of clival chordoma using variable surgical approaches and radiation modalities.

BACKGROUND: Clival chordomas frequently recur because of their location and invasiveness. OBJECTIVE: To investigate clinical, operative, and anatomic factors associated with clival chordoma recurrence. METHODS: Retrospective review of clival chordomas treated at our center from 1993 to 2013. RESULTS: Fifty patients (56% male) with median age of 59 years (range, 8-76) were newly diagnosed with clival chordoma of mean diameter 3.3 cm (range, 1.5-6.7). Symptoms included headaches (38%), diplopia (36%), and dysphagia (14%). Procedures included transsphenoidal (n=34), transoral (n=4), craniotomy (n=5), and staged approaches (n=7). Gross total resection (GTR) rate was 52%, with 83% mean volumetric reduction, values that improved over time. While the lower third of the clivus was the least likely superoinferior zone to contain tumor (upper third=72%/middle third=82%/lower third=42%), it most frequently contained residual tumor (upper third=33%/middle third=38%/lower third=63%; P<.05). Symptom improvement rates were 61% (diplopia) and 53% (headache). Postoperative radiation included proton beam (n=19), cyberknife (n=7), intensity-modulated radiation therapy (n=6), external beam (n=10), and none (n=4). At last follow-up of 47 patients, 23 (49%) remain disease-free or have stable residual tumor. Lower third of clivus progressed most after GTR (upper/mid/lower third=32%/41%/75%). In a multivariate Cox proportional hazards model, male gender (hazard ratio [HR]=1.2/P=.03), subtotal resection (HR=5.0/P=.02), and the preoperative presence of tumor in the middle third (HR=1.2/P=.02) and lower third (HR=1.8/P=.02) of the clivus increased further growth or regrowth, while radiation modality did not. CONCLUSION: Our findings underscore long-standing support for GTR as reducing chordoma recurrence. The lower third of the clivus frequently harbored residual or recurrent tumor, despite staged approaches providing mediolateral (transcranial+endonasal) or superoinferior (endonasal+transoral) breadth. There was no benefit of proton-based over photon-based radiation, contradicting conventional presumptions.

Distinct facilitated diffusion mechanisms by E. coli Type II restriction endonucleases.

The passive search by proteins for particular DNA sequences involving nonspecific DNA is essential for gene regulation, DNA repair, phage defense, and diverse epigenetic processes. Distinct mechanisms contribute to these searches, and it remains unresolved as to which mechanism or blend of mechanisms best suits a particular protein and, more importantly, its biological role. To address this, we compare the translocation properties of two well-studied bacterial restriction endonucleases (ENases), EcoRI and EcoRV. These dimeric, magnesium-dependent enzymes hydrolyze related sites (EcoRI ENase, 5'-GAATTC-3'; EcoRV ENase, 5'-GATATC-3'), leaving overhangs and blunt DNA segments, respectively. Here, we demonstrate that the extensive sliding by EcoRI ENase, involving sliding up to ∼600 bp prior to dissociating from the DNA, contrasts with a larger reliance on hopping mechanism(s) by EcoRV ENase. The mechanism displayed by EcoRI ENase results in a highly thorough search of DNA, whereas the EcoRV ENase mechanism results in an extended, yet less rigorous, interrogation of DNA sequence space. We describe how these mechanistic distinctions are complemented by other aspects of these endonucleases, such as the 10-fold higher in vivo concentrations of EcoRI ENase compared to that of EcoRV ENase. Further, we hypothesize that the highly diverse enzyme arsenal that bacteria employ against foreign DNA involves seemingly similar enzymes that rely on distinct but complementary search mechanisms. Our comparative approach reveals how different proteins utilize distinct site-locating strategies.

DNA looping provides for "intersegmental hopping" by proteins: a mechanism for long-range site localization.

Studies on how transcription factors and DNA modifying enzymes passively locate specific sites on DNA have yet to be reconciled with a sufficient set of mechanisms that can adequately account for the efficiency and speed of this process. This is especially true when considering that these DNA binding/modifying proteins have diverse levels of both cellular copy numbers and genomic recognition site densities. The monomeric bacterial DNA adenine methyltransferase (Dam) is responsible for the rapid methylation of the entire chromosome (with only ~100 Dam copies per cell) and the regulated methylation of closely spaced sites that controls the expression of virulence genes in several human pathogens. Provocatively, we find that Dam travels between its recognition sites most efficiently when those sites are ~500bp apart. We propose that this is manifested by Dam moving between distal regions on the same DNA molecule, which is mediated by DNA looping, a phenomenon we designate as intersegmental hopping. Importantly, an intermediate found in other systems including two simultaneously bound, looped DNA strands is not involved here. Our results suggest that intersegmental hopping contributes to enzymatic processivity (multiple modifications), which invoke recent reports demonstrating that DNA looping can assist in site finding. Intersegmental hopping is possibly used by other sequence-specific DNA binding proteins, such as transcription factors and regulatory proteins, given certain biological context. While a general form of this mechanism is proposed by many research groups, our consideration of DNA looping in the context of processive catalysis provides new mechanistic insights and distinctions.