1 Introduction

Hereditary transthyretin amyloidosis (ATTRv) is an uncommon yet severe, systemic disease resulting from the accumulation of a mutated form of transthyretin (TTR), a physiological transporter of serum thyroxine and retinol-binding protein. Mutations in the TTR gene on chromosome 18 cause protein misfolding, leading to abnormal amyloid fibril deposition in various organs, notably the heart and peripheral nerves. The clinical presentation of ATTRv typically includes progressive sensorimotor polyneuropathy, autonomic dysfunction, and cardiac failure, with a mean survival of 7–10 years after onset if left untreated [1-6].

ATTRv polyneuropathy (ATTRv-PNP) stems from the accumulation of misfolded TTR in the peripheral nervous system (PNS), resulting in length-dependent neuropathy [4, 7]. Pathologically, amyloid fibril deposition initiates in the dorsal root ganglia and autonomic ganglia, where the blood-nerve barrier is more permeable [8]. Subsequently, TTR deposition progresses along the peripheral nerve, following a proximal-to-distal gradient in line with the axoplasmic endoneurial flow [9, 10]. Amyloid fibrils and prefibrillar material predominantly accumulate around endoneurial capillaries, causing extensive microstructural changes and widespread axonal loss within the nerve [11-13].

The typical diagnostic work-up in ATTRv-PNP patients includes a neurological examination coupled with nerve conduction studies (NCS). However, these methods have limitations in detecting early nerve lesions and determining the site and extent of lesions [14]. With various therapies recently approved for ATTRv amyloidosis treatment and others in development, there is an ongoing need for sensitive and reliable biomarkers and imaging markers to aid in disease onset detection, progression tracking, and drug efficacy monitoring [15-23].

Magnetic resonance neurography (MRN) has the potential to overcome diagnostic limitations by directly identifying nerve injuries in several peripheral neuropathies, including ATTRv-PNP [24-31]. Different quantitative MRI parameters, such as apparent T2 relaxation time (T2app), normalized signal intensity (NSI), and spin proton density (ρ), have been shown to detect microstructural changes in peripheral nerves mainly due to an increase in free and unbound water caused by inflammation [32-35]. Diffusion tensor imaging (DTI) and magnetization transfer imaging with the easily computable magnetization transfer ratio (MTR), a semi-quantitative parameter sensitive to myelin changes in various peripheral and central neurological diseases, have recently been shown to identify macromolecular changes and microstructural alterations in nerves of patients affected by ATTRv amyloidosis and other neuropathies [14, 31, 36-42].

In our study, we propose an innovative application of MRN using a dual echo variable echo time (vTE) sequence. Due to its very short echo time (TE), this sequence enables the quantification of the signal from short T2 relaxation time components of the nerve, including myelin and epineurium. It also allows for signal suppression of long T2 relaxation time components in the nerve, such as small unbound water molecules that increase as a consequence of inflammation and endoneurial edema [43].

The higher in-plane spatial resolution of the chosen vTE protocol, when compared to standard MTR and DTI protocols, allows for a better visualization of the different components of the nerve, primarily represented by the inner epineurium and fascicles. This feature enables the extrapolation of morpho-volumetric measurements and helps distinguish the contributions of these different structures to the overall pT2* signal.

Our study aims to evaluate the application of the vTE sequence as a non-invasive method to detect nerve injury in ATTRv patients and to establish a novel potential imaging marker of neuropathy that correlates with disease severity and abnormal results from NCS.

2 Methods

2.3 Image Analysis

For signal quantification, the pseudo-T2* (pT2*) was calculated as an indication of T2*, which is known to be a quantifiable parameter strictly related to the intrinsic T2 relaxation time of a specific tissue, thus reflecting the microstructural composition in vivo, according to the following formula:

$$$$ pseudo-{T_2}^{\ast }=\frac{-\Delta TE}{\ln \left(\frac{\mathrm{iTE}2}{\mathrm{iTE}1}\right)} $$$$

iTE2 and iTE1 are the signal intensities of the second and first echo images, respectively, and ΔTE is the difference between the two echo times [43, 44].

The volume of interest was centered on the right thigh to image the sciatic nerve in the proximal third of the thigh, using a Vitamin E pill placed on the patient's skin as a reference. Figure 1.

For pT2* quantification (ms), the first and last slices of the acquisition were excluded to avoid signal changes due to field inhomogeneities. The central 10 slices were analyzed using ITKSNAP, an open-source segmentation tool. Two neuroradiologists (C.A. and M.P.), each with 6–8 years of experience, blinded to group and clinical stage, manually drew regions of interest (ROIs) around the whole sciatic nerve on each axial slice. The operators then also segmented the endoneurial fascicles and the inner epineurium within the nerve.

Morphometric quantification of sciatic nerve caliber was performed by measuring nerve cross-sectional area (CSA) on each axial slice, using JIM software (Xinapse, version 8). Additionally, volumetric measures of the fascicular area and inner epineurium area were extrapolated (mm²) Figure 2.

The entire segmentation process took approximately 35–40 min per patient. Finally, we calculated the intraclass correlation coefficient (ICC) and the Dice similarity coefficient (DICE) to evaluate inter-operator reliability [47].

3 Results

3.3 pT2* Signal Quantification

Sciatic whole nerve and fascicles pT2* values were significantly higher in patients with ATTRv (whole nerve: mean = 15.65 ± 3.04 ms; fascicles: mean = 18.35 ± 3.48 ms) than in controls (whole nerve: mean = 12.64 ± 3.61 ms; fascicles: mean = 14.74 ± 3.21 ms) in a linear regression analysis, controlling for sex and age (whole nerve: p = 0.009; R = 0.419; fascicles: p = < 0.001; R = 0.541). No significant differences were found in epineurium pT2* values between patients and HC (p = 0.189). Additionally, in the comparison of groups with different degrees of PNP severity (NIS-LL Grade), fascicles pT2* demonstrated differences among the moderate/severe ATTRv-PNP (n = 6), mild ATTRv-PN (n = 12), and HC groups (p = 0.001; F value = 8.397). The post hoc comparison revealed a significant increase in fascicles pT2* values within the moderate/severe PNP group (21.33 ± 3.48 ms) when compared to the control group (14.74 ± 3.21 ms; p = 0.002). A significant difference was also found between mild ATTRv-PNP (17.51 ± 3.09 ms) and HC (p = 0.034) and between moderate/severe and mild ATTRv-PNP (p = 0.042). Figure 3.

The ICC and DICE values were similar for the whole nerve, fascicles, and epineurium segmentation, indicating excellent reliability (ICC: whole nerve = 0.932, fascicles = 0.912, epineurium = 0.878; DICE: whole nerve = 0.928, fascicles = 0.901, epineurium = 0.889) [47].

3.5 Quantitative MRI Parameters Correlate With Clinical Outcomes

We then assessed the role of quantitative vTE sequence as an imaging marker of disease severity by examining its correlation with validated scales of disability (PND-score) and neurologic impairment (NIS and NIS-LL).

A significant association was found between fascicles pT2* values and NIS (R = 0.544 p = 0.030), NIS-LL (R = 0.545 p = 0.032), PND-score (R = 0.592 p < 0.001) and NIS-LL Grade (R = 0.613 p < 0.001), when controlling for sex and age. Figure 4.

3.6 Quantitative MRI Parameters Correlate With NCS Measures

Fascicles pT2* also showed a significant correlation with several electroneurographic parameters. Notably, we observed a negative correlation between fascicles pT2* and CMAP of peroneal (ρ = −0.592 p = 0.026) and tibial nerves (ρ = −0.715 p = 0.006). A negative trend between fascicles pT2* and sural SNAP was also found (ρ = −0.552 p = 0.065).

Patients with moderate to severe PNP also showed a reduction (peroneal nerve: ρ = −0.519 p = 0.042; tibial nerve: ρ = −0.531 p = 0.038) in nerve conduction velocity (NCV), but with a concomitant reduction in CMAP, suggesting a mixed pattern of axonal and demyelinating damage at a later stage of the disease Figure 5.

4 Discussion

In this study, we investigated sciatic nerve involvement in a cohort of ATTRv patients with PNP and showed that the vTE sequence is able to detect pathological changes in the peripheral nerve. In more detail, the main findings of this study are: (i) whole nerve and fascicles pT2* values were increased in ATTRv patients compared to HC, confirming the sensitivity of the vTE sequence to detect pathological microstructural changes within the nerve; (ii) CSA and fascicular area were higher in ATTRv patients compared to HC, whereas the epineurium was only minimally affected; (iii) fascicles pT2* levels reflected disease severity, discriminating mild and moderate/severe PNP from HC; (iv) Lastly, fascicles pT2* correlated with nerve conduction parameters of motor and sensory axonal involvement, appearing as a new promising non-invasive imaging marker sensitive to peripheral neuropathy.

Peripheral nervous system involvement is a primary manifestation of ATTRv amyloidosis. It is characterized by progressive sensorimotor PNP, which substantially impacts patients' quality of life [1-6]. Recently, novel RNA-targeted therapies have demonstrated effectiveness in treating PNP associated with amyloidosis, notably ATTRv amyloidosis, by curtailing abnormal protein formation and stabilizing the normal tetrameric structure of transthyretin [15, 17, 19, 20, 23]. These therapeutic options have remarkably improved the prognosis of ATTRv patients, but timely initiation of treatment remains critical for favorable long-term outcomes [15-23]. In this scenario, it is crucial to establish reliable and feasible imaging markers capable of detecting and quantifying the initial peripheral nerve damage associated with amyloid deposition, thereby enabling early-stage diagnosis and prompt initiation of therapy.

Some previous studies have demonstrated the sensitivity of various quantitative parameters, including NSI, T2app, and proton spin density, in quantifying microstructural changes in the sciatic nerve of ATTRv patients, serving as novel imaging markers for assessing PNS involvement and its progression. Notably, these metrics accurately discriminated patients with ATTRv amyloidosis from HC and had the capability to identify subclinical and early nerve lesions, focusing on the detection of long T2 relaxation time components (unbound water molecules—inflammation) [32-34]. More recently, both MTR and DTI studies have shown promising and unexpected results. In contrast to the well-established axonal pattern observed in ATTRv-PNP, these studies found decreased MTR values and increased RD in DTI analysis within the sciatic nerve of ATTRv patients, both indicating potential myelin involvement [14, 32, 36-42].

Here we report a study using the vTE sequence to assess sciatic nerve injury in symptomatic ATTRv-PNP. The vTE sequence enables visualization of the boundaries of the sciatic nerve and differentiation between the fascicles (hyperintense) and epineurium (hypointense). Consistent with previous studies, we observed that CSA was increased in the patient cohort [32-34]. In this study, we went a step further and extrapolated the fascicular area and the inner epineurium area within the nerve. The increase in CSA and fascicular area observed in patients with overlapping epineurial areas has led us to hypothesize that the larger nerve caliber in the patient group is primarily attributed to an increase in fascicle diameter, while the epineurium appears to remain relatively unaffected [32-34].

ATTRv patients exhibited a higher sciatic whole nerve and fascicles pT2* values compared to healthy controls, highlighting the sensitivity of the vTE sequence in assessing nerve injury. Due to its very short echo time, this sequence enables the detection and quantification of macromolecular content variations within the nerve, especially within the inner epineurium and fascicles. Prefibrillar material and amyloid fibrils, characterized by a short T2 relaxation time, may also be detected using the vTE sequence [43]. Since the epineurium showed no significant involvement and the fascicular area was larger in ATTRv patients, the observed increase in the sciatic nerve pT2* values may be linked to microstructural alterations primarily occurring within the fascicles. These findings align with autopsy and histopathological studies showing that the endoneurium is the most affected inner structure of the nerve in the pathological process of ATTRv-PNP [13]. Higher pT2* values in fascicles of ATTRv patients could be attributed to variations in myelin content or to the accumulation of substances with short-ultrashort T2 times, such as prefibrillar material and amyloid fibrils. However, given the inverse relationship between the content of short T2 time components and pT2* values, amyloid deposition would likely lead to a decrease in pT2*. Therefore, a possible reduction in myelin content in the fascicles as the primary factor driving the observed increase in pT2* cannot be excluded.

These findings are supported by postmortem and biopsy studies wherein areas of myelin loss have been consistently observed in large myelinated fibers, particularly at sites of amyloid deposition. This leads to cytoplasmic degenerative changes within Schwann cells, resulting in a reduction of myelin content in myelinated fibers and of small unmyelinated fibers primarily due to the decreased trophic support [10, 47].

Similar to the other quantitative measures mentioned above, pT2* reflects clinical severity and correlates with all validated clinical outcome measures of ATTRv-PNP (PND-score, NIS, NIS-LL). Interestingly, the vTE sequence can differentiate patients with mild ATTRv-PNP from HC by detecting microstructural changes in the early stages of nerve damage, thus potentially helping the clinical decision in the initiation of therapy.

Being negatively correlated with CMAP values of peroneal and tibial nerves in ATTRv-PNP patients, pT2* appears to reflect the severity and extent of motor axonal loss and/or degeneration. This is clinically relevant because pT2* is measurable and, therefore, applicable for disease monitoring in even more advanced stages of PNP when CMAPs are no longer elicitable [39]. According to previous studies, we also found a reduction in conduction velocity in the later stages of the disease (moderate to severe), confirming the presence of a microstructural pattern distinct from other purely axonal polyneuropathies [32, 39]. If myelin involvement is present, especially in large deep nerves, further investigation using alternative approaches is warranted. It is important to note that the most common misdiagnosis of ATTRv-PNP is chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) [32, 39].

Our study has some limitations. First, the sample size was limited, and we recruited mostly ATTRv patients with mild to moderate neuropathy, while more advanced stages of the disease were underrepresented. Nerve biopsy was not performed in any of the cases, so a radiological-histological correlation was not possible. In addition, the anatomical coverage of the vTE acquisition is relatively small compared to the length of the peripheral nerve and does not allow assessment of differences along the nerve, considering that most peripheral neurological diseases have different patterns of nerve involvement. The sciatic nerve was analyzed in the proximal thigh without further evaluation of tibial and peroneal divisions. Only the right thigh was examined, in line with several previous studies that already focused on one leg, as no significant differences were found between the two legs in ATTRv patients [32, 33, 39]. Additionally, the segmentation of the endoneurial fascicles also included a portion of the inner epineurium, which may have influenced the effective pT2* values extrapolated from the fascicles. Finally, a multi-echo approach, with short-ultrashort TE, may have been more suitable for detecting the microstructural changes that occur within the nerve in relation to amyloid deposition, providing a more precise and sensitive measure of nerve damage.

In conclusion, the vTE MRI sequence proved to be suitable for the detection and quantification of short T2* nerve components in our ATTRv patients, including myelin components, providing a potential imaging marker of nerve damage related to the deposition of aberrant proteins. The derived measure (i.e., pT2*) also reflected disease severity and eventually succeeded in detecting nerve injury during the early disease stages in patients with mild PNP. Longitudinal studies are warranted to assess the role of vTE as a noninvasive, objective, and sensitive imaging marker for the diagnosis and monitoring over time of ATTRv-PNP patients, especially in pre-symptomatic and early symptomatic stages of the disease.

Author Contributions

Carlo Asteggiano: study concept and design, acquisition of data, analysis and interpretation of data, statistical analysis, study supervision and coordination, writing the manuscript. Matteo Paoletti: acquisition of data, analysis and interpretation of data, critical revision of manuscript for intellectual content. Elisa Vegezzi: acquisition of data, analysis and interpretation of data, critical revision of manuscript for intellectual content. Xeni Deligianni: analysis and interpretation of data, critical revision of manuscript for intellectual content. Francesco Santini: analysis and interpretation of data, critical revision of manuscript for intellectual content. Niels Bergsland: analysis and interpretation of data, critical revision of manuscript for intellectual content. Nico Papinutto: analysis and interpretation of data, statistical analysis, critical revision of manuscript for intellectual content. Massimiliano Todisco: acquisition of data, analysis and interpretation of data, critical revision of manuscript for intellectual content. Giuseppe Cosentino: contribution of patients, critical revision of manuscript for intellectual content. Andrea Cortese: study concept and design, acquisition of data, analysis and interpretation of data, critical revision of manuscript for intellectual content. Laura Obici: acquisition of data, critical revision of manuscript for intellectual content. Giovanni Palladini: acquisition of data, critical revision of manuscript for intellectual content. Anna Pichiecchio: study concept and design, acquisition of data, study supervision and coordination, critical revision of manuscript for intellectual content.

Conflicts of Interest

The authors declare no conflicts of interest.