Joel Fontanarosa

A block-based evolutionary optimization strategy to investigate gene-gene interactions in genetic association studies

Multi-locus interactions in genetic association studies are believed to influence the heritability of a number of common diseases. In this study, we propose a discrete optimization strategy to improve the power and computational efficiency of multi-locus association analysis. We implemented an adaptive evolutionary algorithm in combination with a linkage disequilibrium-based discretization approach to reduce the need for exhaustive search for combinations by taking advantage of inherent genomic structure. The method was applied to several simulated disease models as well as to a real genome-wide association study. The results indicate that our method performs as well as or better than the most powerful competing methods for detecting true interactions, and it achieves this performance with improved computational efficiency.

Free Flap Reconstruction Monitoring Techniques and Frequency in the Era of Restricted Resident Work Hours

Importance Free flap reconstruction of the head and neck is routinely performed with success rates around 94% to 99% at most institutions. Despite experience and meticulous technique, there is a small but recognized risk of partial or total flap loss in the postoperative setting. Historically, most microvascular surgeons involve resident house staff in flap monitoring protocols, and programs relied heavily on in-house resident physicians to assure timely intervention for compromised flaps. In 2003, the Accreditation Council for Graduate Medical Education mandated the reduction in the hours a resident could work within a given week. At many institutions this new era of restricted resident duty hours reshaped the protocols used for flap monitoring to adapt to a system with reduced resident labor. Objectives To characterize various techniques and frequencies of free flap monitoring by nurses and resident physicians; and to determine if adapted resident monitoring frequency is associated with flap compromise and outcome. Design, Setting, and Participants This multi-institutional retrospective review included patients undergoing free flap reconstruction to the head and/or neck between January 2005 and January 2015. Consecutive patients were included from different academic institutions or tertiary referral centers to reflect evolving practices. Main Outcomes and Measures Technique, frequency, and personnel for flap monitoring; flap complications; and flap success. Results Overall, 1085 patients (343 women [32%] and 742 men [78%]) from 9 institutions were included. Most patients were placed in the intensive care unit postoperatively (n = 790 [73%]), while the remaining were placed in intermediate care (n = 201 [19%]) or in the surgical ward (n = 94 [7%]). Nurses monitored flaps every hour (q1h) for all patients. Frequency of resident monitoring varied, with 635 patients monitored every 4 hours (q4h), 146 monitored every 8 hours (q8h), and 304 monitored every 12 hours (q12h). Monitoring techniques included physical examination (n = 949 [87%]), handheld external Doppler sonography (n = 739 [68%]), implanted Doppler sonography (n = 333 [31%]), and needle stick (n = 349 [32%]); 105 patients (10%) demonstrated flap compromise, prompting return to the operating room in 96 patients. Of these 96 patients, 46 had complete flap salvage, 22 had partial loss, and 37 had complete loss. The frequency of resident flap checks did not affect the total flap loss rate (q4h, 25 patients [4%]; q8h, 8 patients [6%]; and q12h, 8 patients [3%]). Flap salvage rates for compromised flaps were not statistically different. Conclusions and Relevance Academic centers rely primarily on q1h flap checks by intensive care unit nurses using physical examination and Doppler sonography. Reduced resident monitoring frequency did not alter flap salvage nor flap outcome. These findings suggest that institutions may successfully monitor free flaps with decreased resident burden.

Spatially resolved observation of the spectral hole burning in the Xe(L) amplifier on single and double vacancy 3d → 2p transitions in the 2.62 Å < λ < 2.94 Å range

The analysis of spatially resolved Xe(L) spectra obtained with Z−λ imaging reveals two prominent findings concerning the characteristics of the x-ray amplification occurring in self-trapped plasma channels formed by the focusing of multi-TW subpicosecond 248 nm laser pulses into a high-density gaseous Xe cluster target. They are (1) strongly saturated amplification across both lobes of the Xe(L) hollow atom 3d → 2p emission profile, a breadth that spans a spectral width of ~600 eV, and (2) new evidence for the formation of x-ray spatial modes based on the signature of the transversely observed emission from the narrow trapped zone of the channel. The global characteristics of the spectral measurements, in concert with prior analyses of the strength of the amplification, indicate that the enhancement of the x-ray emission rate by intra-cluster superradiant dynamics plays a leading role in the amplification. This radiative interaction simultaneously promotes (a) a sharp boost in the effective gain, (b) the directly consequent efficient production of coherent Xe(L) x-rays from both single and double vacancy 3d → 2p transition arrays, estimated herein at ~30%, and (c) the development of a very short x-ray pulse width τx. In the limit of sufficiently strong superradiant coupling in the cluster, the system assumes a dynamically collective character and acts as a single homogeneously broadened transition whose effective radiative width approaches the full Xe(L) bandwidth, a breadth that establishes a potential lower limit of τx ~5–10 as, a value substantially less than the canonical atomic time ao/αc 24 as.

Using LASSO regression to detect predictive aggregate effects in genetic studies.

We use least absolute shrinkage and selection operator (LASSO) regression to select genetic markers and phenotypic features that are most informative with respect to a trait of interest. We compare several strategies for applying LASSO methods in risk prediction models, using the Genetic Analysis Workshop 17 exome simulation data consisting of 697 individuals with information on genotypic and phenotypic features (smoking, age, sex) in 5-fold cross-validated fashion. The cross-validated averages of the area under the receiver operating curve range from 0.45 to 0.63 for different strategies using only genotypic markers. The same values are improved to 0.69–0.87 when both genotypic and phenotypic information are used. The ability of the LASSO method to find true causal markers is limited, but the method was able to discover several common variants (e.g., FLT1) under certain conditions.

Comprehensive T-cell immunophenotyping and next-generation sequencing of human papillomavirus (HPV)-positive and HPV-negative head and neck squamous cell carcinomas

The success of programmed cell death 1 (PD‐1) inhibition in achieving a clinical response in a subset of head and neck squamous cell carcinoma (HNSCC) patients emphasizes the need to better understand the immunobiology of HNSCC. Immunophenotyping was performed for 30 HCSCC patients [16 human papillomavirus (HPV)‐positive; 14 HPV‐negative] on matched tissue from the primary tumour site, locally metastatic cervical lymph nodes (LNs), uninvolved local cervical LNs, and peripheral blood. CD4+ and CD8+ T‐cell lymphocytes obtained from tissue were analysed for expression levels of the inhibitory receptors PD‐1, TIM‐3 and CTLA‐4. Next‐generation sequencing of the T‐cell receptor (TCR) β chain was performed on patients (n = 9) to determine receptor repertoire diversity and for clonality analysis. HPV‐negative HNSCC patients, particularly those with stage IV disease, had significantly higher proportions of CD8+ T cells expressing CTLA‐4 in tumour tissue (P = 0.0013) and in peripheral blood (P = 0.0344) than HPV‐positive patients, as well as higher expression levels of TIM‐3+PD‐1+ CD8+ T cells (P = 0.0072) than controls. For all patients, PD‐1 expression on CD8+ T cells – particularly in HPV‐negative HNSCC cases – strongly correlated (r = 0.63, P = 0.013) with tumour size at the primary site. The top CD8+ TCR clones from tumour tissue significantly overlapped with circulating peripheral blood TCR clones (r = 0.946), and HPV‐positive patients had frequently expanded TCR clones that were more hydrophobic – and potentially more immunogenic – than those from HPV‐negative patients. Collectively, our findings demonstrate, for the first time, that high‐stage HPV‐negative HNSCC patients with primary tumours at different sites in the head and neck have elevated peripheral CTLA‐4+CD8+ T‐cell levels, that tumour‐familiar CD8+ T cells are detectable in peripheral blood from HNSCC patients, and that TCRs from HPV‐positive HNSCC patients potentially recognize distinctly immunogenic cognate antigens. However, our findings are preliminary, and need to be further confirmed in a larger patient cohort; also, how these factors affect patient response to immunotherapy needs to be determined. Copyright

Peritumoral Cuffing by T cell Tumor Infiltrating Lymphocytes Distinguishes HPV-Related Oropharyngeal Squamous Cell Carcinoma from Oral Cavity Squamous Cell Carcinoma

BACKGROUND It is unclear why human papillomavirus (HPV)-related head and neck squamous cell carcinoma (HNSCC) has improved clinical behavior compared to HPV-negative HNSCC. We sought to better characterize the immune microenvironment of tongue cancers by examining the CD3 and CD8 TIL pattern in HPV-positive and HPV-negative tumors. METHODS Histologic sections from 40 oral tongue and oropharyngeal cases were analyzed (n=21 HPV DNA-positive, n=19 HPV DNA-negative). CD3 and CD8 T-cell immunostaining were performed on whole-slide sections to quantify tumor-infiltrating lymphocyte (TIL) density and assess its morphology. RESULTS A subset of cases (HPV-positive) displayed a unique TIL pattern consisting of circumferential peritumoral population T cells, which was absent in the HPV-negative cases. The presence of peritumoral cuffing was strongly predictive of improved recurrence-free survival compared to cases that lacked this morphologic pattern of immune infiltrate. Four HPV-positive cases lacked the pattern, including two cases with disease recurrence. CONCLUSIONS For the first time, we show an architectural pattern of immune infiltrate in HNSCC is seen exclusively in HPV-positive patients with improved recurrence-free survival and suggests an organized host immunological response contributes to disease control.

Observation of a curve crossing mechanism in the field ionization of inner-shell excited single Xe 33+(2p̄) and Double Xe 34+(2s̄2p̄) vacancy states

Anomalous intensity dependent quenching of the Xe(L) hollow atom emission from the single Xe³³⁺(2p̅) and double Xe³⁴⁺(2s̅2p̅) vacancy 3d→2p transition arrays indicates the presence of a state-selective curve crossing mechanism for the field ionization of the excited 3d⁴ configuration by intense (2-3 × 10²⁰ × W/cm²) 248-nm radiation.

Observation of a Curve Crossing Mechanism in the Field Ionization of Inner-Shell Excited Single

Anomalous intensity dependent quenching of the Xe(L) hollow atom emission from the single Xe³³⁺(2p̅) and double Xe³⁴⁺(2s̅2p̅) vacancy 3d→2p transition arrays indicates the presence of a state-selective curve crossing mechanism for the field ionization of the excited 3d⁴ configuration by intense (2-3 × 10²⁰ × W/cm²) 248-nm radiation.

{\rm Xe}^{33+}{(2\bar{\rm p})}

Anomalous intensity dependent quenching of the Xe(L) hollow atom emission from the single Xe³³⁺(2p̅) and double Xe³⁴⁺(2s̅2p̅) vacancy 3d→2p transition arrays indicates the presence of a state-selective curve crossing mechanism for the field ionization of the excited 3d⁴ configuration by intense (2-3 × 10²⁰ × W/cm²) 248-nm radiation.

and Double

The analysis of gene-gene interactions related to common complex human diseases is complicated by the increasing scale of genetic association analysis. Concurrent with the advances in genetic technology that led to these large data sets, improvements have been made in parallel computing with graphics processing units (GPUs). The data-intensive nature of genetic association analysis makes this problem particularly suitable for improved computation with the powerful computing resources available in GPUs. In this study, we present a GPU-accelerated discrete optimization strategy to improve the computational efficiency of multi-locus association analysis. We implemented an adaptive evolutionary algorithm that takes advantage of linkage disequilibrium to reduce the need for exhaustive search for combinations of genetic markers. The proposed GPU algorithm was shown to have improved efficiency and equivalent power relative to the CPU version.