Amyotrophic Lateral Sclerosis (ALS) is a progressive neurological disorder that entails degeneration of both upper and lower motor neurons. The primary motor cortex (M1) in patients with Upper Motor Neuron (UMN) impairment is pronouncedly hypointense in Magnetic Resonance (MR) T2* contrast, as a consequence of iron deposits caused by neuroinflammatory reaction and cortical microgliosis. In the present study, 3D gradient-recalled multi-echo sequences were used on a 7 Tesla MR system to acquire T2*-weighted images and Quantitative Susceptibility Mapping (QSM) targeting M1 at high spatial resolution in patients with ALS and in healthy controls.

17 patients with ALS (limb onset) and 13 HC participated to this study, with their understanding and written consent in accordance with the protocol approved by the competent Ethics Commitee and in compliance with national legislation. The UMN impairment of ALS patients was evaluated for each individual limb using a composite semiquantitative score (UMN-score) ranging 0~8^[1].

Data were collected with a 7T MR950 (GE Healthcare) system equipped with a 2ch-tx/32ch-rx head coil (NovaMedical). The protocol included a 3D gradient-recalled multi-echo sequence with TR=54.1ms, TEs=5.6ms, 12ms, 18.3ms, 24.7ms, 31.1ms, 37.5ms, 43.9ms, spatial resolution of 0.5×0.5×1mm³. T2*-weighted images were obtained by averaging the magnitude data obtained from each individual echo.

In the right hemisphere of HC, the following four ROIs were drawn, covering the full-thickness cortex of: M1 (hand knob), primary somatosensory cortex (S1), superior parietal lobule, anterior cingulate gyrus. In each hemisphere of both ALS patients and HC, two ROIs were drawn to delineate the deep cortical layers of M1 corresponding to Penfield's areas of the hand (hM1, in the “hand knob”) and foot (fM1, in the paracentral lobule). One additional ROI delineated the splenium of the corpus callosum (SCC).

χ maps were computed from the real and imaginary data of each echo with the iLSQR method^[2], and one final resultant χ map was generated by averaging the χ maps obtained from each individual echo^[3]. χ was expressed in parts per billion [ppb] with respect to the average χ in the SCC, in every single subject.

In HC, the predicted content of iron (ρ, expressed in mg/100g of tissue) as a function of age was calculated in M1, S1, non-S1 parietal cortex and prefrontal cortex^[4].

The correlation between χ and iron content n HC's cortex was evaluated with Spearman rank test. A Mann-Whitney test was used to compare average χ in HC’s M1 with χ in the M1 subregion corresponding to the clinically most affected limb of ALS patients, on the basis of their UMN-score. Correlations between χ in the M1 subregion corresponding to the clinically most affected limb of ALS patients and UMN-scores of the corresponding limbs were measured by Spearman rank test.

χ in the ROIs of four cortical regions (M1, S1, non-S1 parietal cortex and prefrontal cortex) strongly correlated with the expected content of non-haemin iron in brain tissues in HC: r=0.71; p<0.0001 (Figure 1). The linear function that best fitted the relationship between χ and ρ was χ=21.79×ρ−67.22, and the goodness of the linear fit was R²=0.56.

χ in M1 was significantly higher (p<0.028) in ALS patients than in HC (Figure 2). In HC, the average χ across the four M1 subregions (deep cortical layers in hM1 and fM1, left and right) was 37±12ppb, while in ALS patients, in the M1 subregions corresponding to their most impaired limb according to the UMN-score, χ was 52±18ppb.

χ in patients’ M1 significantly correlated (r=0.53; p<0.03) with the UMN-score of the corresponding limb (Figure 3). The linear function that best fitted the relationship between χ and UMN-score was UMN-score=0.102×χ–2.37, and the goodness of the linear fit was R²=0.37.

Based on the linear relation between χ and the expected iron concentration in different cortical regions in HC (Figure 2), our in vivo results agree with previous ex vivo findings, which demonstrated the iron accumulation in microglial cells infiltrated in the deep layers of M1^[5]. One recent study conducted on a 3T MRI system observed an increase in χ in the M1 of ALS patients^[6] however the specificity of such increase to the cortical deep layers and its relation with the clinical condition was not reported. In fact, our results show that χ in M1 significantly correlates with UMN impairment, rather than being a candidate direct biomarker of ALS. QSM might be proved useful in measuring M1 iron concentration, as a possible in vivo biomarker of UMN burden and neuroinflammation in ALS patients.