Utility of multiparametric MRI including T1/T2 mapping and IVIM/diffusion imaging for the evaluation of non-obstructive azoospermia

Abstract

Introduction and objectives: The management of non-obstructive azoospermia (NOA) remains challenging because no predictive test for the presence of localized spermatogenesis exists. Previous work considered MRI techniques, such as spectroscopy (MRS) and diffusion weighted imaging (DWI), in this role. We report here data from a prospective study evaluating additional advanced MRI sequences for predicting spermatogenesis in patients with NOA.

Methods: 9 fertile volunteers and 18 men with NOA were prospectively recruited. Each participant underwent a novel multi-parametric MRI consisting of T1 and T2 mapping as well as intravoxel incoherent motion (IVIM) and diffusion weighted imaging (DWI). A single radiologist drew representative regions of interest on the best quality images for each sequence and recorded the mean values. Sperm extraction procedure results were recorded. Two-end points were evaluated: NOA versus fertile controls and the presence of viable sperm within the NOA cohort. The data were analyzed per patient. Nonparametric and logistic regression statistical analysis were used.

Results: 9 fertile men (median 43 years old, 2 children) and 18 men with NOA (median 37 years old, 0 children) were studied. 11 of the 18 men with NOA had testicle sampling. 4 men with NOA had viable sperm. Follicle-stimulating hormone and testosterone levels were not significantly different among NOAmen with and without sperm (p-value = 0.58 and 0.25). Nonparametric analysis with the Wilcoxon rank sum test showed T2 relaxation time was lower among NOA patients (median 101 vs 135 ms, p-value = 0.002), apparent diffusion coefficient (ADC) was higher among NOA patients (median 127.9 vs. 106.7 × 10^-5 mm²/sec, p-value = 0.005). T1 relaxation time, alpha (Water diffusion heterogeneity index), D (IVIM-based apparent diffusion coefficient), DDC (Distributed diffusion coefficient) and D* (pseudodiffusion) were also significantly different. On logistic regression analysis, both T2 and ADC were associated with NOA; The odds of NOA decreased by 6% for each msec increase in T2 (p-value = 0.02) while the odds of NOA increased by 11% for each 10⁻⁵ mm²/sec increase in ADC, (p-value = 0.02). T2 yielded a larger area under the receiver operating characteristic curve than ADC (0.87 versus 0.84). Alpha, D, DDC and D* also predicted NOA. Amongst men with NOA who underwent testicle sampling, T2 was lower in testicles of patients with no sperm retrieved (median 73 vs 134. msec, p-value = 0.02). The remaining variables were not significantly different between the cohorts.

Conclusions: In spite of the small sample size, particularly for men with NOA who underwent sperm extraction, these results suggest that several novel MRI parameters, such as T2 relaxation time and certain IVIM/DWI parameters, are able to distinguish between fertile men and men with NOA as well as potentially predict successful sperm extraction in men with NOA. Additional larger prospective studies of men with NOA undergoing sperm extraction are warranted.

Keywords: Infertility; MRI; Nonobstructive azoospermia.