Noninvasive Prediction of Sperm Retrieval Using Diffusion Tensor Imaging in Patients with Nonobstructive Azoospermia

Sikang Gao, Jun Yang, Dong Chen, Xiangde Min, Chanyuan Fan, Peipei Zhang, Qiuxia Wang, Zhen Li, Wei CaiJournal of Imaging 2023 · DOI

TL;DR

This study explores Diffusion Tensor Imaging (DTI) as a non-invasive method to predict the success of sperm retrieval in men with nonobstructive azoospermia (NOA). They found that Mean Diffusivity (MD) values in the testes can effectively differentiate between patients who will and will not successfully retrieve sperm, potentially avoiding unnecessary invasive procedures.

What Problem Does This Paper Address?

Male infertility is a significant issue, and azoospermia, the absence of sperm in ejaculate, is a severe form. Nonobstructive azoospermia (NOA) accounts for a large proportion of azoospermia cases and is characterized by a dysfunction in sperm production within the testes. The primary treatment for NOA is microdissection testicular sperm extraction (mTESE), a surgical procedure. However, the success rate of mTESE is only around 43-63%, meaning a substantial number of patients undergo this invasive surgery without retrieving sperm. Current predictors for mTESE success, such as age, testicular volume, and hormone levels, are unreliable. This leaves a critical need for a non-invasive, accurate method to predict sperm retrieval outcomes before surgery, thereby reducing patient burden and healthcare costs.

Methods

The researchers utilized Diffusion Tensor Imaging (DTI), a specialized MRI technique that maps the diffusion of water molecules within tissues. DTI provides quantitative parameters like Mean Diffusivity (MD) and Fractional Anisotropy (FA). MD measures the overall extent of water diffusion, while FA quantifies the directionality of this diffusion, reflecting tissue microstructure.

Seventy patients diagnosed with NOA were enrolled. Scrotal MRI scans were performed, and DTI parameters (MD and FA) were extracted from regions of interest (ROIs) placed on the testicular parenchyma. These DTI values were then compared between two groups: those who successfully retrieved sperm via mTESE (Group A) and those who did not (Group B). Statistical analyses, including ROC analysis, were used to evaluate the diagnostic performance of these DTI parameters in predicting sperm retrieval success.

Datasets

The study utilized data from 70 patients diagnosed with NOA. These patients were recruited from Tongji Hospital between September 2017 and March 2023. The dataset was divided into two groups based on the outcome of their mTESE procedure:

  • Group A (Successful Sperm Retrieval): 29 patients.
  • Group B (Failed Sperm Retrieval): 41 patients.

No new benchmark datasets were introduced; the study relied on the collected patient data for its analysis.

Key Insights and Ideas

  • MD as a Biomarker: The most significant finding is that Mean Diffusivity (MD) in testicular tissue appears to be a strong indicator of sperm retrieval success. Higher MD values were observed in patients who successfully retrieved sperm. This suggests that changes in testicular tissue density and cellularity, reflected by water diffusion, are directly linked to the presence of viable sperm-producing cells.
  • FA's Limited Role: Interestingly, Fractional Anisotropy (FA), which measures the directional diffusion of water, did not show a significant difference between the successful and failed retrieval groups. This implies that while the overall diffusion of water is informative, the specific directional organization of testicular structures, as measured by FA, might not be as critical for predicting sperm retrieval in NOA.
  • Hormonal Predictors Fall Short: The study confirmed that standard sex hormone levels (FSH, LH, estradiol, testosterone, prolactin) did not significantly differ between the successful and failed sperm retrieval groups. This reinforces the limitations of current hormonal assessments in predicting mTESE outcomes.
  • Non-Invasive Potential: The core idea is to leverage DTI's non-invasive nature to avoid unnecessary surgeries. By identifying patients with a high likelihood of failed sperm retrieval through DTI, clinicians can potentially spare them the risks, costs, and emotional toll of mTESE.

Key Results

  • MD Difference: Mean Diffusivity (MD) values were significantly higher in the successful sperm retrieval group (Group A) compared to the failed sperm retrieval group (Group B) (p < 0.001).
  • ROC Analysis: The Area Under the Curve (AUC) for MD in predicting sperm retrieval was 0.865. At a cutoff value of 1609.8 × 10⁻⁶ mm²/s, MD achieved a sensitivity of 72.2% and a specificity of 97.5%.
  • FA No Difference: No statistically significant difference was found in Fractional Anisotropy (FA) values between the two groups (p > 0.05).
  • Hormone Levels: Sex hormone levels showed no statistically significant differences between the successful and failed sperm retrieval groups.

Limitations and Future Work

  • Single-Center Study: The study was conducted at a single center with a relatively small sample size (70 patients). This limits the generalizability of the findings.
  • FA Prediction: The study suggests that FA might not be a reliable predictor for sperm retrieval success in NOA patients, aligning with previous research. However, further investigation is needed to definitively determine its role.
  • Future Directions:
    • Multicenter Validation: Future research should involve larger, multicenter studies to validate these findings and confirm the diagnostic accuracy of DTI.
  • Automated Approaches: Exploring automated classification and segmentation techniques (e.g., using convolutional neural networks) for DTI analysis could improve efficiency and consistency.
  • Combined Modalities: Investigating the use of DTI in combination with other imaging techniques or clinical factors could lead to more comprehensive and accurate prediction models.
  • Broader DTI Parameters: Further exploration of other DTI parameters and varying imaging protocols (e.g., different diffusion directions, b-values) might reveal additional insights.

Summary updated

Abstract

Microdissection testicular sperm extraction (mTESE) is the first-line treatment plan for nonobstructive azoospermia (NOA). However, studies reported that the overall sperm retrieval rate (SRR) was 43% to 63% among men with NOA, implying that nearly half of the patients fail sperm retrieval. This study aimed to evaluate the diagnostic performance of parameters derived from diffusion tensor imaging (DTI) in predicting SRR in patients with NOA. Seventy patients diagnosed with NOA were enrolled and classified into two groups based on the outcome of sperm retrieval during mTESE: success (29 patients) and failure (41 patients). Scrotal magnetic resonance imaging was performed, and the DTI parameters, including mean diffusivity and fractional anisotropy, were analyzed between groups. The results showed that there was a significant difference in mean diffusivity values between the two groups, and the area under the curve for mean diffusivity was calculated as 0.865, with a sensitivity of 72.2% and a specificity of 97.5%. No statistically significant difference was observed in fractional anisotropy values and sex hormone levels between the two groups. This study demonstrated that the mean diffusivity value might serve as a useful noninvasive imaging marker for predicting the SRR of NOA patients undergoing mTESE.
BibTeX
@article{gao2023noninvasive,
  author = {Sikang Gao and Jun Yang and Dong Chen and Xiangde Min and Chanyuan Fan and Peipei Zhang and Qiuxia Wang and Zhen Li and Wei Cai},
  title = {{Noninvasive Prediction of Sperm Retrieval Using Diffusion Tensor Imaging in Patients with Nonobstructive Azoospermia}},
  journal = {Journal of Imaging},
  year = {2023},
  doi = {10.3390/jimaging9090182},
}
Figure