Therneau, Terry M.

ObjectiveMutations in Charcot–Marie–Tooth disease (CMT) genes are the cause of rare familial forms of polyneuropathy. Whether allelic variability in CMT genes is also associated with common forms of polyneuropathy—considered “acquired” in medical parlance—is unknown. Chemotherapy‐induced peripheral neuropathy (CIPN) occurs commonly in cancer patients and is individually unpredictable. We used CIPN as a clinical model to investigate the association of non‐CMT polyneuropathy with CMT genes.MethodsA total of 269 neurologically asymptomatic cancer patients were enrolled in the clinical trial Alliance N08C1 to receive the neurotoxic drug paclitaxel, while undergoing prospective assessments for polyneuropathy. Forty‐nine CMT genes were analyzed by targeted massively parallel sequencing of genomic DNA from patient blood.ResultsA total of 119 (of 269) patients were identified from the 2 ends of the polyneuropathy phenotype distribution: patients that were most and least susceptible to paclitaxel polyneuropathy. The CMT gene PRX was found to be deleteriously mutated in patients who were susceptible to CIPN but not in controls (p = 8 × 10−3). Genetic variation in another CMT gene, ARHGEF10, was highly significantly associated with CIPN (p = 5 × 10−4). Three nonsynonymous recurrent single nucleotide variants contributed to the ARHGEF10 signal: rs9657362, rs2294039, and rs17683288. Of these, rs9657362 had the strongest effect (odds ratio = 4.8, p = 4 × 10−4).InterpretationThe results reveal an association of CMT gene allelic variability with susceptibility to CIPN. The findings raise the possibility that other acquired polyneuropathies may also be codetermined by genetic etiological factors, of which some may be related to genes already known to cause the phenotypically related Mendelian disorders of CMT. Ann Neurol 2014;76:727–737

Abstract Introduction: Symptom control is an important objective in the care of patients with cancer and a critical goal for long-term survivors of antineoplastic treatments. Commonly used chemotherapies, including paclitaxel, are associated with a toxic effect on the peripheral nervous system in a substantial fraction of patients, termed chemotherapy induced peripheral neuropathy (CIPN), that is often debilitating. CIPN cannot be predicted from clinical parameters in neurologically normal subjects, suggesting a genetic basis as the root cause. Patients with hereditary neuropathy (HN) due to one of ~50 recognized genes are prone to developing severe CIPN even if the hereditary predisposition is clinically silent. Assessment of CIPN has usually been an afterthought of cancer treatment studies using crude symptom measurement instruments. N08C1 and N08CA are recent clinical trials_the latter analyzed only in early 2013_focusing on extensive phenotyping by patient reported outcomes (PRO) to support genetic investigations of paclitaxel-related CIPN. Massively parallel (“nextgen”) DNA sequencing is the genetic technology of choice and represents an as-yet unrealized potential to study cancer symptom control. Methods: Exome sequencing_analysis of 20,794 genes at nucleotide resolution_was performed on blood of CIPN cases (n=73) and controls (n=46) from N08C1. The primary endpoint was the association of HN genes with CIPN. Bioinformatic analysis determined SNV and other genetic variations across 6.2x107 nucleotides. The association of genes with CIPN was statistically examined by the over-dispersion test C-alpha. The role of singleton SNV was assessed by a burden test. N08CA served as validation. Results: 386,644 SNV were detected with an estimated sensitivity of 90% and specificity of 99.9%; the orthogonal validation rate (PCR of >1000 genotypes) was 99.7%. The association of the minor allele (C>T) at rs7349683 in the gene EPHA5, previously reported by Baldwin et al. (2012) for CALGB40101, was confirmed in N08C1. Two hereditary neuropathy genes were identified as new CIPN candidate genes: Firstly, ARHGEF10 was highly significant with p=5x10-4 based on single gene testing and p<0.02 after Bonferroni correction for multiple testing of 41 hereditary neuropathy genes with non-synonymous SNV. Three SNV were predictive, rs17683288, rs2294039 and rs9657362, the latter with an odds ratio (OR)=3.62. Validation in an independent patient group (n=75) from another paclitaxel CIPN trial, N08CA, confirmed the impact of ARHGEF10 on CIPN (p<0.025; rs9657362 OR=2.51). Secondly, peraxin (PRX) was disrupted by non-synonymous singleton SNV in n=8 cases and n=0 controls (OR=6.4; unadjusted p=0.04). 7 of the 8 mutations were also scored as deleterious by both Polyphen 2 and SIFT. Because PRX SNV were non-recurrent validation was performed in N08C1 by Sanger sequencing. Several neuro-genes across the exome were noted as additional CIPN gene candidates. Conclusions: ARHGEF10 and PRX associations with CIPN suggest that CIPN and HN might share genetic roots in a subset of patients. Construct validity of N08C1 is supported by confirmation of a published result from CALGB40101 relating EPHA5 to CIPN. Germline exome sequencing in cancer patients is highly informative for understanding complex pharmacogenomic traits including outcomes in symptom control research. Citation Format: Andreas S. Beutler, Amit Kulkarni, Rahul Kanwar, Rui Qin, Julie M. Cunningham, Terry M. Therneau, Charles L. Loprinzi. Sequencing symptom control: Results from the Alliance N08C1 and N08CA genetics of chemotherapy neuropathy trials. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr LB-196. doi:10.1158/1538-7445.AM2013-LB-196

ObjectiveTo examine trends in glucocorticoid (GC) use and dosing among patients diagnosed with rheumatoid arthritis (RA) over time.MethodsA population‐based inception cohort of RA patients diagnosed during 1980–2007 was followed longitudinally through their medical records until death, migration, or December 31, 2008. GC start and stop dates were collected, along with doses in prednisone equivalents.ResultsThe study population comprised 349 patients (68% women) diagnosed in 1980–1994 and 464 (69% women) diagnosed in 1995–2007, with a median followup of 15.3 and 5.7 years, respectively. A higher proportion of patients started GCs in their first year of disease in 1995–2007 (68% versus 36%; P < 0.001), but the starting dose (mean 8.7 versus 10.3 mg; P = 0.08) and cumulative dose in the first year of use (mean 1.8g [mean daily dose 4.9 mg] versus 2.1 gm [mean daily dose 5.8 mg]; P = 0.48) were not different. A higher proportion also discontinued GCs in their first year of disease in the 1995–2007 cohort (P < 0.001). These differences in GC initiation and discontinuation persisted throughout followup. Prevalence of GC use was higher in the 1995–2007 cohort for the first 3 years of disease.ConclusionMore patients are starting GCs early in their disease course now compared to previously, which is consistent with established treatment guidelines. A higher proportion are also discontinuing GCs, but the proportion of patients taking GCs at any given point of disease during the first 4 years is higher now than previously. Despite early addition of a disease‐modifying antirheumatic drug, some patients may not be able to discontinue GCs over the long term.

Abstract Summary: Mate pair library sequencing is an effective and economical method for detecting genomic structural variants and chromosomal abnormalities. Unfortunately, the mapping and alignment of mate-pair read pairs to a reference genome is a challenging and time-consuming process for most next-generation sequencing alignment programs. Large insert sizes, introduction of library preparation protocol artifacts (biotin junction reads, paired-end read contamination, chimeras, etc.) and presence of structural variant breakpoints within reads increase mapping and alignment complexity. We describe an algorithm that is up to 20 times faster and 25% more accurate than popular next-generation sequencing alignment programs when processing mate pair sequencing. Availability: http://bioinformaticstools.mayo.edu/research/bima/ Contact: vasm@mayo.edu or vasmatzis.george@mayo.edu Supplementary Information: Supplementary data are available at Bioinformatics online.

Abstract Background: Given the high technical reproducibility and orders of magnitude greater resolution than microarrays, next-generation sequencing of mRNA (RNA-Seq) is quickly becoming the de facto standard for measuring levels of gene expression in biological experiments. Two important questions must be taken into consideration when designing a particular experiment, namely, 1) how deep does one need to sequence? and, 2) how many biological replicates are necessary to observe a significant change in expression? Results: Based on the gene expression distributions from 127 RNA-Seq experiments, we find evidence that 91% ± 4% of all annotated genes are sequenced at a frequency of 0.1 times per million bases mapped, regardless of sample source. Based on this observation, and combining this information with other parameters such as biological variation and technical variation that we empirically estimate from our large datasets, we developed a model to estimate the statistical power needed to identify differentially expressed genes from RNA-Seq experiments. Conclusions: Our results provide a needed reference for ensuring RNA-Seq gene expression studies are conducted with the optimally sample size, power, and sequencing depth. We also make available both R code and an Excel worksheet for investigators to calculate for their own experiments.

Abstract Motivation: Batch effects are due to probe-specific systematic variation between groups of samples (batches) resulting from experimental features that are not of biological interest. Principal component analysis (PCA) is commonly used as a visual tool to determine whether batch effects exist after applying a global normalization method. However, PCA yields linear combinations of the variables that contribute maximum variance and thus will not necessarily detect batch effects if they are not the largest source of variability in the data. Results: We present an extension of PCA to quantify the existence of batch effects, called guided PCA (gPCA). We describe a test statistic that uses gPCA to test whether a batch effect exists. We apply our proposed test statistic derived using gPCA to simulated data and to two copy number variation case studies: the first study consisted of 614 samples from a breast cancer family study using Illumina Human 660 bead-chip arrays, whereas the second case study consisted of 703 samples from a family blood pressure study that used Affymetrix SNP Array 6.0. We demonstrate that our statistic has good statistical properties and is able to identify significant batch effects in two copy number variation case studies. Conclusion: We developed a new statistic that uses gPCA to identify whether batch effects exist in high-throughput genomic data. Although our examples pertain to copy number data, gPCA is general and can be used on other data types as well. Availability and implementation: The gPCA R package (Available via CRAN) provides functionality and data to perform the methods in this article. Contact: reesese@vcu.edu or eckel@mayo.edu Supplementary information: Supplementary data are available at Bioinformatics online.

Current methods used to assess calibration are limited, particularly in the assessment of prognostic models. Methods for testing and visualizing calibration (e.g. the Hosmer–Lemeshow test and calibration slope) have been well thought out in the binary regression setting. However, extension of these methods to Cox models is less well known and could be improved. We describe a model-based framework for the assessment of calibration in the binary setting that provides natural extensions to the survival data setting. We show that Poisson regression models can be used to easily assess calibration in prognostic models. In addition, we show that a calibration test suggested for use in survival data has poor performance. Finally, we apply these methods to the problem of external validation of a risk score developed for the general population when assessed in a special patient population (i.e. patients with particular comorbidities, such as rheumatoid arthritis).

ObjectiveTo determine whether the impact of aging on cardiovascular disease (CVD) risk in patients with rheumatoid arthritis (RA) differs from that in the general population (as estimated by the Framingham Risk Score [FRS]).MethodsA population‐based inception cohort of Olmsted County, Minnesota residents ages ≥30 years who fulfilled the American College of Rheumatology 1987 criteria for RA in 1988–2008 was assembled and followed up until death, migration, or July 1, 2012. Data on CVD events were collected by medical records review. The 10‐year FRS for CVD was calculated. Cox models adjusted for FRS were used to examine the influence of age on CVD risk.ResultsThe study included 563 patients with RA without prior CVD (mean age 55 years, 72% women, and 69% seropositive [i.e., rheumatoid factor and/or anti–citrullinated protein antibody positive]). During a mean followup of 8.2 years, 98 patients developed CVD (74 seropositive and 24 seronegative), but the FRS predicted only 59.7 events (35.4 seropositive and 24.3 seronegative). The gap between observed and predicted CVD risk increased exponentially across age, and the effect of age on CVD risk in seropositive RA was nearly double its effect in the general population, with additional log(age) coefficients of 2.91 for women (P = 0.002) and 2.06 for men (P = 0.027).ConclusionOur findings indicate that age exerts an exponentially increasing effect on CVD risk in seropositive RA, but no increased effect among seronegative patients. The causes of accelerated aging in patients with seropositive RA deserve further investigation.

ObjectiveTo study left ventricular (LV) geometry in patients with rheumatoid arthritis (RA) and no history of heart failure compared with that in subjects with neither RA nor a history of heart failure, and to determine the impact of RA on LV remodeling.MethodsA cross‐sectional, community‐based study was conducted among adult (age ≥50 years) patients with RA and age‐ and sex‐matched subjects with neither RA nor a history of heart failure. All participants underwent standard 2‐dimensional Doppler echocardiography. LV geometry was classified into the following 4 categories based on relative wall thickness and sex‐specific cutoffs for the LV mass index: concentric remodeling, concentric hypertrophy, eccentric hypertrophy, or normal geometry.ResultsAmong 200 patients with RA and 600 age‐ and sex‐matched subjects without RA, the mean age was 65 years, and 74% of the individuals in both cohorts were female. Compared with subjects without RA, patients with RA were significantly more likely to have abnormal LV geometry (odds ratio [OR] 1.44, 95% confidence interval [95% CI] 1.03–2.00), even after adjusting for cardiovascular risk factors and comorbidities. Among subjects with abnormal LV geometry, the odds of concentric LV remodeling were significantly increased in patients with RA (OR 4.73, 95% CI 2.85–7.83). In linear regression analyses, the LV mass index appeared to be lower in patients with RA who were currently receiving corticosteroids (β ± SE −0.082 ± 0.027, P = 0.002), even after adjusting for cardiovascular risk factors and comorbidities.ConclusionRA was strongly associated with abnormal LV remodeling (particularly concentric LV remodeling) among RA patients without heart failure. This association remained significant after adjustment for cardiovascular risk factors and comorbidities. RA disease–related factors may promote changes in LV geometry. The biologic mechanisms underlying LV remodeling warrant further investigation.