Structure dictates the function of many RNAs, but secondary RNA structure analysis is either labor intensive and costly or relies on computational predictions that are often inaccurate. These limitations are alleviated by integration of structure probing data into prediction algorithms. However, existing algorithms are optimized for a specific type of probing data. Recently, new chemistries combined with advances in sequencing have facilitated structure probing at unprecedented scale and sensitivity. These novel technologies and anticipated wealth of data highlight a need for algorithms that readily accommodate more complex and diverse input sources. We implemented and investigated a recently outlined probabilistic framework for RNA secondary structure prediction and extended it to accommodate further refinement of structural information. This framework utilizes direct likelihood-based calculations of pseudo-energy terms per considered structural context and can readily accommodate diverse data types and complex data dependencies. We use real data in conjunction with simulations to evaluate performances of several implementations and to show that proper integration of structural contexts can lead to improvements. Our tests also reveal discrepancies between real data and simulations, which we show can be alleviated by refined modeling. We then propose statistical preprocessing approaches to standardize data interpretation and integration into such a generic framework. We further systematically quantify the information content of data subsets, demonstrating that high reactivities are major drivers of SHAPE-directed predictions and that better understanding of less informative reactivities is key to further improvements. Finally, we provide evidence for the adaptive capability of our framework using mock probe simulations.

RNA SHAPE experiments have become important and successful sources of information for RNA structure prediction. In such experiments, chemical reagents are used to probe RNA backbone flexibility at the nucleotide level, which in turn provides information on base pairing and therefore secondary structure. Little is known, however, about the statistics of such SHAPE data. In this work, we explore different representations of noise in SHAPE data and propose a statistically sound framework for extracting reliable reactivity information from multiple SHAPE replicates. Our analyses of RNA SHAPE experiments underscore that a normal noise model is not adequate to represent their data. We propose instead a log-normal representation of noise and discuss its relevance. Under this assumption, we observe that processing simulated SHAPE data by directly averaging different replicates leads to bias. Such bias can be reduced by analyzing the data following a log transformation, either by log-averaging or Kalman filtering. Application of Kalman filtering has the additional advantage that a prior on the nucleotide reactivities can be introduced. We show that the performance of Kalman filtering is then directly dependent on the quality of that prior. We conclude the paper with guidelines on signal processing of RNA SHAPE data.

Abstract

INTRODUCTION

Mutations in isocitrate dehydrogenase (IDH) have been investigated as a prognostic biomarker in glioma. The presence of the IDH mutation (IDHm) is associated with 2-hydroxyglutarate (2HG) production and inhibition of glutamate synthesis (McBrayer, Cell 2018). Hyperpolarized carbon-13 (HP-13C) MRI enables dynamic measurements of in-vivo metabolism using a [2-13C]pyruvate labeled probe that undergoes conversion to [2-13C]lactate and [5-13C]glutamate. Here, we present HP [2-13C]pyruvate data from healthy volunteers and patients with IDHm diffuse glioma. Due to its intrinsic low signal-to-noise ratio (SNR), we demonstrate the ability of post-processing denoising to improve its utility and aid in detection of metabolic changes associated with IDHm.

METHODS

Dynamic HP 13C data were acquired following intravenous injection of [2-13C]pyruvate from five healthy volunteers and one patient with IDHm grade III astrocytoma. A novel multi-resolution frequency specific multislice EPI sequence was used to obtain [2-13C]pyruvate, [5-13C]glutamate, and downfield and upfield [2-13C]lactate signals (3s temporal resolution, pyruvate/lactate/glutamate spatial resolutions = 0.75x0.75cm2/ 2.25x2.25cm2/ 2.25x2.25cm2, 5 slices 3cm thick). Following phase correction, patch-based tensor decomposition denoising was applied to metabolite images. Metabolite differences between normal-appearing white matter (NAWM) and T2 lesion were examined for the patient data.

RESULTS

HP [2-13C]pyruvate imaging is able to simultaneously probe glycolytic ([2-13C]lactate) and oxidative ([5-13C]glutamate) metabolism. Denoised pyruvate/lactate/glutamate signals achieved a 4-9/3-6/3-7 fold increase in SNR. T2 lesion exhibited decreased glutamate-to-pyruvate and glutamate-to-lactate AUC ratios versus contralateral NAWM (p< 0.018, p < 1.5e-5), consistent with IDH mutant status.

CONCLUSION

We successfully demonstrated the feasibility of applying variable resolution HP [2-13C]pyruvate metabolic imaging to detect IDHm specific metabolism. This technique addresses a major hurdle in HP 13C MRI by improving SNR while permitting robust metabolism quantification. Future studies will optimize methods for acquiring and processing data to evaluate further data acquired from IDHm glioma patients. Supported by NIH T32 CA151022, P01 CA118816, and NICO.

Purpose

To investigate high-resolution hyperpolarized (HP) ¹³ C pyruvate MRI for measuring cerebral perfusion in the human brain.

Methods

HP [1-¹³ C]pyruvate MRI was acquired in five healthy volunteers with a multi-resolution EPI sequence with 7.5 × 7.5 mm² resolution for pyruvate. Perfusion parameters were calculated from pyruvate MRI using block-circulant singular value decomposition and compared to relative cerebral blood flow calculated from arterial spin labeling (ASL). To examine regional perfusion patterns, correlations between pyruvate and ASL perfusion were performed for whole brain, gray matter, and white matter voxels.

Results

High resolution 7.5 × 7.5 mm² pyruvate images were used to obtain relative cerebral blood flow (rCBF) values that were significantly positively correlated with ASL rCBF values (r = 0.48, 0.20, 0.28 for whole brain, gray matter, and white matter voxels respectively). Whole brain voxels exhibited the highest correlation between pyruvate and ASL perfusion, and there were distinct regional patterns of relatively high ASL and low pyruvate normalized rCBF found across subjects.

Conclusion

Acquiring HP ¹³ C pyruvate metabolic images at higher resolution allows for finer spatial delineation of brain structures and can be used to obtain cerebral perfusion parameters. Pyruvate perfusion parameters were positively correlated to proton ASL perfusion values, indicating a relationship between the two perfusion measures. This HP ¹³ C study demonstrated that hyperpolarized pyruvate MRI can assess cerebral metabolism and perfusion within the same study.

This study aimed to implement a multimodal ¹H/HP-¹³C imaging protocol to augment the serial monitoring of patients with glioma, while simultaneously pursuing methods for improving the robustness of HP-¹³C metabolic data. A total of 100 ¹H/HP [1-¹³C]-pyruvate MR examinations (104 HP-¹³C datasets) were acquired from 42 patients according to the comprehensive multimodal glioma imaging protocol. Serial data coverage, accuracy of frequency reference, and acquisition delay were evaluated using a mixed-effects model to account for multiple exams per patient. Serial atlas-based HP-¹³C MRI demonstrated consistency in volumetric coverage measured by inter-exam dice coefficients (0.977 ± 0.008, mean ± SD; four patients/11 exams). The atlas-derived prescription provided significantly improved data quality compared to manually prescribed acquisitions (n = 26/78; p = 0.04). The water-based method for referencing [1-¹³C]-pyruvate center frequency significantly reduced off-resonance excitation relative to the coil-embedded [¹³C]-urea phantom (4.1 ± 3.7 Hz vs. 9.9 ± 10.7 Hz; p = 0.0007). Significantly improved capture of tracer inflow was achieved with the 2-s versus 5-s HP-¹³C MRI acquisition delay (p = 0.007). This study demonstrated the implementation of a comprehensive multimodal ¹H/HP-¹³C MR protocol emphasizing the monitoring of steady-state/dynamic metabolism in patients with glioma.

Background

Real-time metabolic conversion of intravenously-injected hyperpolarized [1-¹³C]pyruvate to [1-¹³C]lactate and [¹³C]bicarbonate in the brain can be measured using dynamic hyperpolarized carbon-13 (HP-¹³C) MRI. However, voxel-wise evaluation of metabolism in patients with glioma is challenged by the limited signal-to-noise ratio (SNR) of downstream ¹³C metabolites, especially within lesions. The purpose of this study was to evaluate the ability of higher-order singular value decomposition (HOSVD) denoising methods to enhance dynamic HP [1-¹³C]pyruvate MRI data acquired from patients with glioma.

Methods

Dynamic HP-¹³C MRI were acquired from 14 patients with glioma. The effects of two HOSVD denoising techniques, tensor rank truncation-image enhancement (TRI) and global-local HOSVD (GL-HOSVD), on the SNR and kinetic modeling were analyzed in [1-¹³C]lactate data with simulated noise that matched the levels of [¹³C]bicarbonate signals. Both methods were then evaluated in patient data based on their ability to improve [1-¹³C]pyruvate, [1-¹³C]lactate and [¹³C]bicarbonate SNR. The effects of denoising on voxel-wise kinetic modeling of k_PL and k_PB was also evaluated. The number of voxels with reliable kinetic modeling of pyruvate-to-lactate (k_PL) and pyruvate-to-bicarbonate (k_PB) conversion rates within regions of interest (ROIs) before and after denoising was then compared.

Results

Both denoising methods improved metabolite SNR and regional signal coverage. In patient data, the average increase in peak dynamic metabolite SNR was 2-fold using TRI and 4-5 folds using GL-HOSVD denoising compared to acquired data. Denoising reduced k_PL modeling errors from a native average of 23% to 16% (TRI) and 15% (GL-HOSVD); and k_PB error from 42% to 34% (TRI) and 37% (GL-HOSVD) (values were averaged voxelwise over all datasets). In contrast-enhancing lesions, the average number of voxels demonstrating within-tolerance k_PL modeling error relative to the total voxels increased from 48% in the original data to 84% (TRI) and 90% (GL-HOSVD), while the number of voxels showing within-tolerance k_PB modeling error increased from 0% to 15% (TRI) and 8% (GL-HOSVD).

Conclusion

Post-processing denoising methods significantly improved the SNR of dynamic HP-¹³C imaging data, resulting in a greater number of voxels satisfying minimum SNR criteria and maximum kinetic modeling errors in tumor lesions. This enhancement can aid in the voxel-wise analysis of HP-¹³C data and thereby improve monitoring of metabolic changes in patients with glioma following treatment.

Background

Dynamic hyperpolarized (HP)-¹³C MRI has enabled real-time, non-invasive assessment of Warburg-related metabolic dysregulation in glioma using a [1-¹³C]pyruvate tracer that undergoes conversion to [1-¹³C]lactate and [¹³C]bicarbonate. Using a multi-parametric ¹H/HP-¹³C imaging approach, we investigated dynamic and steady-state metabolism, together with physiological parameters, in high-grade gliomas to characterize active tumor.

Methods

Multi-parametric ¹H/HP-¹³C MRI data were acquired from fifteen patients with progressive/treatment-naïve glioblastoma [prog/TN GBM, IDH-wildtype (n = 11)], progressive astrocytoma, IDH-mutant, grade 4 (G4A^IDH+, n = 2) and GBM manifesting treatment effects (n = 2). Voxel-wise regional analysis of the cohort with prog/TN GBM assessed imaging heterogeneity across contrast-enhancing/non-enhancing lesions (CEL/NEL) and normal-appearing white matter (NAWM) using a mixed effects model. To enable cross-nucleus parameter association, normalized perfusion, diffusion, and dynamic/steady-state (HP-¹³C/spectroscopic) metabolic data were collectively examined at the ¹³C resolution. Prog/TN GBM were similarly compared against progressive G4A^IDH+ and treatment effects.

Results

Regional analysis of Prog/TN GBM metabolism revealed statistically significant heterogeneity in ¹H choline-to-N-acetylaspartate index (CNI)_max, [1-¹³C]lactate, modified [1-¹³C]lactate-to-[1-¹³C]pyruvate ratio (CEL_val > NEL_val > NAWM_val); [1-¹³C]lactate-to-[¹³C]bicarbonate ratio (CEL_val > NEL_val/NAWM_val); and ¹H-lactate (CEL_val/NEL_val > NAWM_undetected). Significant associations were found between normalized perfusion (cerebral blood volume, nCBV; peak height, nPH) and levels of [1-¹³C]pyruvate and [1-¹³C]lactate, as well as between CNI_max and levels of [1-¹³C]pyruvate, [1-¹³C]lactate and modified ratio. GBM, by comparison to G4A^IDH+, displayed lower perfusion %-recovery and modeled rate constants for [1-¹³C]pyruvate-to-[1-¹³C]lactate conversion (k_PL), and higher ¹H-lactate and [1-¹³C]pyruvate levels, while having higher nCBV, %-recovery, k_PL, [1-¹³C]pyruvate-to-[1-¹³C]lactate and modified ratios relative to treatment effects.

Conclusions

GBM consistently displayed aberrant, Warburg-related metabolism and regional heterogeneity detectable by novel HP-¹³C/¹H imaging techniques.