Diffusion tensor imaging metrics in diagnosis of HTLV‐1‐associated myelopathy

Abstract

Objective

In the diagnosis of HTLV‐1‐associated myelopathy (HAM), while magnetic resonance imaging (MRI) is essential to exclude other diseases, its power is limited regarding HAM diagnosis, as only 30% of affected patients present with spinal cord atrophy. Diffusion tensor imaging (DTI) may enable the detection of damage in the white matter microstructure. Here, we quantitatively assess spinal cord damage using DTI and evaluate conventional MRI parameters of the spinal cord in HTLV‐1‐infected individuals.

Methods

This cross‐sectional study involved 33 HTLV‐1 carriers, 28 patients with definite‐HAM, and 11 seronegative healthy subjects (HS). Region‐of‐interest (ROI)‐based fractional anisotropy (FA) and mean diffusivity (MD) measurements were performed in the upper thoracic and lumbar regions of the spinal cord. Thoracic index was defined as 1/ (anteroposterior diameter × transverse diameter) measured at the fifth 5th vertebral level. Receiver operating characteristic (ROC) curve analysis was used to determine optimal cutoff FA, MD, and thoracic index values.

Results

Spinal cord atrophy was observed in 15 (53.6%) patients with definite‐HAM. The area under the ROC curve in the thoracic spinal cord was 0.824 (95% CI, 0.716–0.932), 0.839 (95% CI: 0.736–0.942), and 0.838 (95% CI: 0.728–0.949) for FA, MD, and the thoracic index, respectively. Lower FA and higher MD values were observed in the definite‐HAM group compared to HTLV‐1 carriers and HS at the T5 vertebral level (p < 0.01).

Interpretation

Complementary to conventional MRI, DTI analysis of the spinal cord and thoracic index determination can offer additional insight that may prove useful in the diagnosis of HAM.

Introduction

Human T‐cell leukemia virus type 1 (HTLV‐1) infection has afflicted 10 million people worldwide. ¹ HTLV‐1‐associated morbidity is highly underestimated: While only 5% of infected individuals develop HTLV‐1‐associated myelopathy (HAM) or adult T‐cell leukemia (ATL), ² more than 50% of those not diagnosed with these conditions suffer from other related diseases or syndromes, including HTLV‐1‐associated arthropathy, sicca syndrome, bladder dysfunction, erectile dysfunction, and infective dermatitis. ³ , ⁴ , ⁵ Definite HAM is characterized by slowly progressive spastic paraparesis, which may initially present unilaterally, but invariably develops bilaterally, being associated with generalized hyperreflexia and extensor plantar reflex. ⁶ Approximately 20% of HTLV‐1‐infected individuals without definite HAM may present bladder dysfunction, characterized by urinary urgency, incontinence, and/or nocturia; in the absence of motor disability, these patients are classified as having probable HAM. ⁴ , ⁵ , ⁶ Cerebrospinal fluid analysis and magnetic resonance imaging (MRI) of the spinal cord reveal no unique characteristics in patients with definite HAM, yet both are essential to exclude other diseases. Depending on the stage of disease, MRI may reveal an edematous, normal or atrophic spinal cord, with or without signal alterations. ⁶ Neuropathological findings indicate that atrophy of the spinal cord seems to mainly occur in the anteroposterior direction. ⁷ , ⁸ Most pathological alterations in HAM have been reported in the thoracolumbar cord. ⁹ , ¹⁰ Spinal cord atrophy can be radiologically quantified ⁸ , ⁹ , ¹¹ , ¹² and is the most common finding observed on conventional MRI. However, atrophy usually appears only years after the onset of disease. ¹³ Moreover, the spinal cord is normal on MRI in up to 30% of patients presenting neurological findings compatible with definite HAM. ¹²

Diffusion tensor imaging (DTI) combined with fiber tractography allows for the mapping and consequent evaluation of different tracts of white matter in the central nervous system. ¹⁴ , ¹⁵ The DTI technique measures the diffusion of water molecules, describing movement in a three‐dimensional space, thereby mapping directionality. ¹⁶ The degree of anisotropy can be quantified by some measures, such as fractional anisotropy (FA) and mean diffusivity (MD). ¹⁷ , ¹⁸ DTI‐tractography of the spinal cord has been successfully used to evaluate subtle pathological abnormalities in cervical spondylotic myelopathy, ¹⁹ monitor disease activity in patients with multiple sclerosis ²⁰ and intramedullary tumors, ²¹ , ²² , ²³ and has been suggested as a possible tool to detect disease progression in patients with amyotrophic lateral sclerosis, ²⁴ as well as to evaluate functional motor ability in patients with acute transverse myelitis. ²⁵ The present study endeavored to quantitatively assess spinal cord alterations in HTLV‐1‐infected subjects using DTI and measures of anteroposterior and transverse diameters of the spinal cord.

Material and Methods

Study participants

The present cross‐sectional study recruited 61 HTLV‐1‐infected subjects followed from 2 to 14 years at the Multidisciplinary HTLV‐1 Clinic of the Prof. Edgard Santos University Hospital Complex, located in Salvador, Bahia‐Brazil. All study subjects underwent recent neurological evaluations in 2020. HTLV‐1 infection was diagnosed by positivity on serological enzyme‐linked immunosorbent assay (ELISA) (Cambridge Biotech Corp., Worcester, MA, USA), with confirmation provided by western blot (HTLV blot, Genelab, Singapore). Of the 61 recruited subjects, 33 were HTLV‐1 carriers and 28 patients were diagnosed with definite HAM; 11 seronegative healthy subjects (HS) were also included. This study was approved by the Institutional Review Board of the Federal University of Bahia. All included participants freely agreed to participate and provided a written term of informed consent.

Case definition

The studied subjects were classified as either HTLV‐1 carriers, that is, HTLV‐1‐infected individuals without neurological signs or symptoms, or infected patients with definite HAM. ²⁶ Briefly, definite HAM is a non‐remitting progressive spastic paraparesis in which sensory, urinary and anal sphincter signs or symptoms may or may not be present. All definite HAM cases presented an Osame motor disability score (OMDS) greater than 1, and other disorders that could mimic HAM were excluded. Seronegative controls were healthy workers who tested seronegative for HTLV‐1 and did not present any signs or symptoms of neurological disease.

Study design

All subjects answered a standardized questionnaire and underwent neurological physical examination by two independent neurologists; proviral load was also quantitatively determined. Two scales were used to assess neurological disability in individuals infected with HTLV‐1: The extended disability status scale (EDSS) ²⁷ and OMDS. ²⁸ Individuals additionally underwent a 10‐m walking test as previously described. ²⁹ The outcomes of interest were the presence of spinal cord atrophy, thoracic index quantification, and changes in diffusion tensor tractography metrics, that is, FA and MD. Exclusion criteria consisted of any history or signs of unrelated diseases with the potential to affect the central nervous system, spine radiation therapy, previous spine surgery, contraindication to MR imaging, and relevant motion artifacts.

HTLV‐1 proviral load

DNA was extracted from 10⁶ peripheral blood mononuclear cells (PBMC) using proteinase K and the salting‐out method. HTLV‐1 proviral load was quantified by real‐time PCR as previously described using an ABI Prism 7700 Sequence detector system (Applied Biosystems), ³⁰ with albumin DNA used as an endogenous reference. Normalized HTLV‐1 proviral load values were calculated as the ratio of HTLV‐1 DNA average copy number/albumin DNA average copy number × 2 × 10, ⁶ and expressed as the number of HTLV‐1 copies per 10⁶ PBMCs.

MR examination

Conventional MRI with DTI was performed on all subjects using a 1.5‐Tesla‐MR scanner (Siemens; Symphony) with a 16‐channel spinal coil. The protocol consisted of sagittal T2‐weighted images (turbo spin echo sequence TE/TR = 131/5010 ms, 3.0 mm slice thickness, FOV = 320 mm), SPACE 3D sequence TE/TR 124/1500 ms, measured voxel size = 1.0 x 1.0 x 1.0 mm, FOV = 256 mm, and transversal T2‐weighted sequence TE/TR 95/4990 ms, FOV = 170 mm) at the 5th thoracic vertebral level and the lower cord immediately above the conus medullaris. The 5th thoracic vertebral level was chosen to reflect the mid‐thoracic spinal cord level. Diffusion weighting was applied in 12 gradient directions with a b‐value = 0 and 1000s/mm² in the axial plane at the same levels, TE/TR 124/1500 ms, measured voxel size = 2.7 × 1.6 × 4.5 mm, FOV = 200 mm. The overall scanning time for the two DTI acquisitions was 1m49s. DTI data were analyzed using syngo.via imaging software (Siemens Healthcare GmbH, Erlangen‐Germany). On axial images, individual regions of interest (ROI) were drawn freehand to circumscribe the hemicord, encompassing white and gray matter (Fig. 1). Due to the smaller size of the spinal cord cross‐section analyzed, it was not possible to obtain valid measurements for separate more minute ROIs. The anteroposterior and transversal diameters were respectively defined as the distance from the ventral to the dorsal border and from the left to right border of the spinal cord. Spinal cord MRIs were examined for the presence of diffuse atrophy. The thoracic and lumbar index were calculated using the following equation: 1/ anteroposterior diameter (AP) x transverse diameter (TD), that is, higher values indicated greater degrees of spinal cord atrophy. Independent assessments of spinal cord images and FA and MD measurements were conducted by a neuroradiologist and a neurologist with experience in neuroimaging; disagreements were settled by consensus. All evaluators were blinded to the clinical status of the subjects.

Figure 1.

Region‐of‐interest (ROI) placement. Placement of the measurement ROIs on the axial B0 image at the 5th thoracic vertebral level (A) and the lower cord immediately above the conus medullaris (B).

Statistical analysis

A database was built using REDCap, an internet‐based software platform developed at Vanderbilt University. ³¹ Data were tested for normality using the Kolmogorov–Smirnoff test. Comparisons among categorical variables were made using Pearson's Chi‐squared test or Fisher's exact test. Comparisons among DTI parameters and volumetric results were performed between the three groups using the Kruskal–Wallis test. Receiver operating characteristic (ROC) curve analysis considering the area under the curve (AUC) was employed to compare the performance of thoracic index, FA and MD measurements in predicting HAM. Spearman's rank correlation coefficients were calculated to evaluate correlations between thoracic index values, and DTI and clinical parameters. All statistical analyses were carried out using SPSS version 22 software; results were considered significant when p < 0.05.

Results

Patient demographic characteristics, duration of follow‐up, the presence of associated diseases and proviral load determinations are listed in Table 1. No differences were seen regarding the demographic features of the three participant groups, nor with respect to time of follow‐up. Median HTLV‐1 proviral load was higher in definite‐HAM patients than in HTLV‐1 carriers (p < 0.02).

Table 1.

Demographic features, associated disease status and proviral load in HTLV‐1‐infected subjects and seronegative individuals.

Characteristics	HAM (n = 28)	HTLV‐1 carriers (n = 33)	Seronegative (n = 11)	p value
Female sex n (%)	19 (67.9%)	19 (57.6%)	6 (54.5%)	0.63 1
Age (years)	53.86 (±12.16)	52.7 (±11)	52.2 (±14.38)	0.89 2
Follow‐up (years)	5.5 (2–8.75)	11 (5–14)	–	0.87 3
Sicca syndrome	0	0	–	–
HTLV‐1 associated arthropathy	5 (17.8%)	5 (15.1%)	–	>0.99 4
Diabetes	6 (21.4%)	2 (6.1%)	–	0.12 4
HBV infection	0	0	–	–
HCV infection	1 (3.6%)	1 (3%)	–	>0.99 4
Proviral load 5 (copies/106 PBMC)	93,893 (49,408–166,835)	20,692 (6,216–82,434)	–	0.12 3

HAM, HTLV‐1‐associated myelopathy; HBV, Hepatitis B virus; HCV, Hepatitis C virus.

¹ χ ² test.

² Kruskal–Wallis test.

³ Mann–Whitney U test.

⁴ Fisher’s exact test.

⁵ Median [interquartile range].

MRI inspection revealed atrophy of the spinal cord in 15 (53.6%) patients with definite‐HAM in contrast to none of the HTLV‐1 carriers or HS (p < 0.001). The thoracic area of the spine, as well as the anteroposterior and transverse diameters measured at T5, were inferior in the group of patients with definite HAM compared to HTLV‐1 carriers and seronegative individuals (p < 0.001, Kruskal–Wallis test). Higher thoracic index values were observed in the definite HAM group (p < 0.001, Mann–Whitney U test) (Table 2). DTI measurements of the spinal cord are shown in Table 3. FA values at the 5th thoracic vertebral level were reduced in the definite HAM group (p < 0.001) along with considerably elevated MD values (p < 0.001) (Table 3). Thoracic and lumbar spinal cord parameters did not differ significantly among HTLV‐1 carriers and seronegative individuals.

Table 2.

Frequency of spinal cord atrophy and MRI measurements of the spinal cord in HTLV‐1‐infected subjects and seronegative individuals.

	HAM (n = 28)	HTLV‐1 carriers (n = 33)	Seronegative (n = 11)	p value
Spinal atrophy 1 n (%)	15 (53.6%)	0	0	<0.001 2
Vertebral level: T5
Thoracic area (cm²) 3	0.260 (0.220–0.320)	0.360 (0.330–0.390)	0.400 (0.387–0.442)	<0.001 4
Anteroposterior diameter (cm) 3	0.480 (0.450–0.525)	0.540 (0.520–0.620)	0.590 (0.525–0.645)	<0.001 4
Transverse diameter (cm) 3	0.730 (0.665–0.785)	0.820 (0.780–0.880)	0.870 (0.825–0.915)	<0.001 4
Thoracic index (1/cm²) 5	2.76 (2.52–3.43)	2.19 (2.01–2.46)	2.02 (1.74–2.19)	<0.001 4
Vertebral level: T12‐L1
Lumbar area (cm²) 3	0.550 (0.455–0.625)	0.580 (0.530–0.650)	0.675 (0.547–0.725)	0.01 4
Anteroposterior diameter (cm) 3	0.690 (0.475–0.780)	0.730 (0.650–0.770)	0.760 (0.715–0.807)	0.06 4
Transverse diameter (cm) 3	0.900 (0.815–0.935)	0.910 (0.850–0.960)	0.920 (0.880–1.012)	0.13 4
Lumbar index (1/cm²) 5	1.59 (1.37–1.94)	1.53 (1.37–1.76)	1.39 (1.27–1.58)	0.06 4

HAM, HTLV‐1‐associated myelopathy.

¹ On inspection.

² χ ² test.

³ Median (interquartile range)

⁴ Kruskal–Wallis test.

⁵ 1/(anteroposterior diameter × transverse diameter).

Table 3.

Diffusion tensor imaging parameters in the spinal cords of HTLV‐1 infected subjects and seronegative individuals.

Vertebral level	DTI parameters	HAM (n = 26) 1	HTLV‐1 carriers (n = 31)	Seronegative (n = 10)	p value 2
T5	Fractional anisotropy	0.519 (0.434–0.565)	0.642 (0.554–0.707)3	0.688 (0.574–0.704)3	<0.001
	Mean diffusivity	1.521 (1.409–1.736)	1.259 (1.135–1.425)3	1.228 (1.084–1.408)3	<0.001
T12‐L1	Fractional anisotropy	0.433 (0.382–0.523)	0.482 (0.399–0.547)	0.545 (0.508–0.621)3	0.015
	Mean diffusivity	1.140 (1.064–1.230)	0.963 (0.892–1.166)3	1.116 (0.921–1.349)	0.037

Measurements presented as median values (interquartile range). DTI, diffusion tensor imaging; HAM, HTLV‐1‐associated myelopathy.

Level of significance for difference between definite‐HAM versus carriers and versus seronegative is p < 0.01 (³) by Mann‐Whitney‐U‐Test.

¹ Due to artifacts, images were only analyzed in 26 of 28 patients with HAM.

² Kruskal–Wallis test.

ROC curve analysis using a cutoff FA value of 0.545 at the 5th thoracic vertebral level successfully discriminated definite HAM subjects with 83.9% sensitivity, while the rate of false HAM patient detection (1‐specificity) was 26.9% (AUC 0.824; 95% CI: 0.716–0.932). The optimal cutoff value established for MD was 1.3445 x 10⁻³ mm²/s, providing a sensitivity of 74.2% and specificity of 84.6% regarding the detection of definite‐HAM patients (AUC 0.839; 95% CI: 0.736–0.942). ROC curve analysis also demonstrated the capability of the thoracic index assessment to discriminate between HTLV‐1 carriers and definite‐HAM patients using a cutoff value of 2.48 (AUC 0.838; 95% CI: 0.728–0.949), achieving 84.4% sensitivity, and 80.8% specificity (Figure 2). Lumbar spinal cord parameters were not found to discriminate between the HTLV‐1‐infected groups. The rate of HAM patients correctly classified using MD values (≥ cutoff) was 84.6% (22/26) compared to 73% (19/26) using FA. In addition, the rate of false identification of HAM (carriers with thoracic index ≥, or FA value ≤ cutoff) was 18.7% (Table 4).

Figure 2.

ROC curve analysis of fractional anisotropy (FA), mean diffusivity (MD), and thoracic index (TI) values in 31 asymptomatic HTLV‐1 carriers and 26 HAM patients at the T5 vertebral level. MD: area under the curve: 0.839 (95% CI: 0.736–0.942), optimal cutoff: 1.3445, with 74.2% sensitivity and 84.6% specificity at cutoff. FA: area under the curve: 0.824 (95% CI: 0.716–0.932), optimal cutoff 0.545, 83.9% sensitivity, and 73.1% specificity at cutoff. TI: area under the curve: 0.838 (95% CI: 0.728–0.949), optimal cutoff 2.48, 84.4% sensitivity and 80.8% specificity at cutoff.

Table 4.

MRI findings in HTLV‐1‐infected subjects and seronegative individuals, stratified by DTI parameter and thoracic index cutoff values.

	HAM N = 26 1	HTLV‐1 carriers N = 31	Seronegative n (%) N = 11
Spinal atrophy 2 present	N = 14	0	0
Thoracic index ≥2.48	14 (100)	0	0
Fractional anisotropy (T5) ≤0.545	14 (100)	0	0
Mean diffusivity (T5) ≥1.3445 × 10−3 mm²/sec	13 (92.8)	0	0
Normal spine 2	N = 12	N = 31	N = 11
Thoracic index ≥2.48	8 (66.6)	6 (18.7)	1 (9)
Fractional anisotropy (T5) ≤0.545	5 (41.6)	6 (18.7)	2 (18.2)
Mean diffusivity (T5) ≥1.3445 × 10−3 mm²/sec	9 (75)	9 (28.1)	3 (27.2)

¹ Due to artifacts, images were only analyzed in 26 of 28 patients with definite HAM.

² On inspection.

We did not find a correlation between illness duration and DTI parameters or the thoracic index, but there was a direct correlation between 10‐m walk test and lumbar MD values (r = 0.519, p = 0.03). Clinical disability scores, expressed as median (interquartile range) values on the EDSS and OMDS were 5.5 (4–6.5) and 5 (3–6.75) in the HAM group, respectively. At the 5th thoracic vertebral level, EDSS and OMDS scores were found to strongly correlate with the following parameters: spinal cord area, anteroposterior and transverse diameters, thoracic index, FA, and MD. A negative correlation was noted between FA and thoracic index, while the former correlated positively with area, AP and TD parameters. A direct correlation was observed between MD and thoracic index, while negative correlations were detected between MD and the AP, TD, and area parameters (Table 5).

Table 5.

Correlations between DTI parameters, imaging findings and disability scale scores in HTLV‐1 carriers and HAM patients at the 5th thoracic vertebral level.

	DTI parameters		Disability scales
	FA	MD	EDSS	OMDS
Area	0.67 (<0.001)	−0.54 (<0.001)	−0.64 (<0.001)	−0.66 (<0.001)
AP	0.56 (<0.001)	−0.45 (<0.001)	−0.54 (<0.001)	−0.58 (<0.001)
TD	0.55 (<0.001)	−0.46 (<0.001)	−0.49 (<0.001)	−0.58 (<0.001)
Thoracic index	−0.65 (<0.001)	0.54 (<0.001)	0.60 (<0.001)	0.62 (<0.001)
EDSS	−0.58 (<0.001)	0.59 (<0.001)
OMDS	−0.60 (<0.001)	0.57 (<0.001)

Spearman correlations were calculated between parameters. The ‘r’ indexes are shown (p value). FA, fractional anisotropy; MD, mean diffusivity; AP, anteroposterior diameter; TD, transverse diameter; thoracic index, 1/AP × TD; EDSS, Extended Disability Status; OMDS, Osame Motor Disability Score.

Discussion

HTLV‐1 was the first retrovirus identified. ³² However, as it remains a neglected infection, there are no complementary exams that assist in the diagnosis of HAM, and no effective antiviral drugs have been developed to treat HTLV‐1. Conventional MRI has failed to demonstrate atrophy or other findings related to this disease in 33.4% to 76.5% of cases meeting the clinical criteria of definite HAM. ⁸ , ¹⁰ , ¹² , ³³ , ³⁴ , ³⁵ Moreover, while conventional MRI may detect abnormalities in definite HAM, a more highly sensitive technique is required to investigate spinal cord involvement in individuals who do not present clinical disease, as well as to determine whether the spinal cord injury detectable on imaging may bear correlations with illness duration and disease severity. DTI metrics are based on measurements of the three‐dimensional diffusivity of water, which is directed by different “barriers”, such as axons, cell membranes and particularly myelin sheaths. ³⁶ Pathological processes impairing the microstructure of central nervous system tissue, for example, the diffuse loss of myelin and axons evidenced by neuropathological studies in definite‐HAM patients, alter the anisotropy of water diffusion. ⁷ , ³⁶ Although previous studies have documented FA abnormalities in patients with HAM¹¹, the sensitivity of DTI parameters in the diagnosis of definite HAM has not been determined to date.

The definite‐HAM and HTLV‐1 carriers involved in the present study presented similar demographic and epidemiological profiles. As expected, proviral load was elevated in patients with HAM compared to HTLV‐1 carriers. HAM development has been associated with increased HTLV‐1 proviral load and the female sex, ⁶ which explains the higher proportion of females in the definite‐HAM group. The most common and nonspecific finding observed on MRI of the thoracic spinal cord is atrophy. ¹³ On inspection, thoracic spinal cord atrophy was observed in 53.6% of the definite‐HAM patients evaluated herein, which stands in agreement with previous findings. ¹⁰ , ¹² , ³⁴ This greater sensitivity of the thoracic index when compared to binary categorization of atrophy reflects lower discrimination and accuracy of visual inspection alone by a radiologist.

In the spinal cord, inflammatory myelitis is characterized by decreased FA values in the region of T2‐weighted lesions, with increased FA values seen at lesion boundaries. ¹⁵ Vilchez et al., ¹¹ in addition to conventional MRI of the cervical cord and thoracolumbar spine, investigated DTI parameters in the evaluation of patients with definite HAM and HTLV‐1 carriers. These authors identified significant correlations exclusively between reduced FA in dorsal fiber and neurological findings. In the present study, the definite‐HAM group showed significantly decreased FA and increased mean diffusivity values in the thoracic spine compared to HS and HTLV‐1 carriers. The decreased FA values may be related to reduce amount of extracellular water due to cytotoxic edema, axonal cluster regeneration or cellular infiltration by inflammatory cells. ¹⁵ Increased MD values could be related to larger extracellular spaces (due to axonal and neuronal loss) in combination with spinal cord atrophy. ³⁷ , ³⁸

The main pathological feature characterizing HAM is the presence of chronic inflammation with mononuclear cell infiltration and diffuse degeneration of both myelin and axons, notably in the thoracic cord. ⁷ The pattern of DTI metrics found in patients with HAM is similar to the one observed in other infectious or inflammatory myelopathies as multiple sclerosis, but DTI parameters may rule out other neurologic disorders of the spinal cord as ischemic myelopathy and acute injury due to compression. ¹⁵ , ²⁰ , ³⁹ Regarding the diagnosis of HAM, the DTI metrics seems a promising supplementary diagnostic tool in addition to the clinical criteria and liquor. Moreover, early signs of HTLV‐1 medullary injury not observed in the conventional MRI may be detectable. The present results suggest that increased MD and decreased FA in the thoracic cord of HAM patients may reflect axonal loss, demyelination and parenchymal inflammatory infiltration, which precisely correlate with previously reported histological findings. ⁷ , ⁴⁰ , ⁴¹ Nonetheless, this remains a speculative hypothesis, as animal studies examining diffusion parameters and corresponding histology would be required to confirm or refute this premise.

In contrast to our findings, some previous reports do not demonstrate clear relationships between the severity of disease and spinal cord parameters. ⁸ , ¹¹ , ¹² , ⁴² However, a study by Liu et al. (2014) did find that spinal cord atrophy was correlated with ambulatory index. ⁹ Moreover, our results showed that clinical disability scores correlated with spinal cord area, anteroposterior and transverse diameters and the thoracic index at the 5th thoracic vertebral level. With regard to DTI parameters, FA and MD in the thoracic cord were also found to correlate with disease severity, in contrast to previous findings. ¹¹ This could be explained by the low number of patients with definite HAM in previous investigations, as well as the use of the American spinal injury association impairment (ASIA) scale to assess the effects of spinal cord injury. Unlike EDSS, neurological disability evaluations made using the ASIA scale do not account for bowel or bladder dysfunction.

Our study has several limitations. The number of seronegative controls included was relatively small, that is, by increasing the number of these individuals, differences in DTI derivative measures might approach significance between carriers and seronegative controls. As only a limited number of patients who presented probable HAM underwent MRI, comparisons between these patients and those with definite HAM were not performed. Although patients with probable HAM present no motor disability, these individuals should be evaluated in future studies, as abnormalities in DTI values were found to directly correlate with the EDSS; notably, patients with probable HAM present EDSS scores ≥2. ⁴ In addition, the present investigation employed a 1.5‐T MRI system with a 16‐channel spinal coil, which offers inferior image quality compared to 3.0 T systems; moreover, ROIs were defined manually and diffusion weighting was applied in 12 gradient directions, which may affect the measurement of DTI parameters. Despite these limitations, the thoracic index proved to be a simple and useful tool in the diagnosis of HAM, providing excellent accuracy (83%) to discriminate between HTLV‐1 carriers and HAM patients.

The well‐described histopathologic findings detailing damage to the spinal cord caused by the inflammatory response in HTLV‐1‐infected subjects may serve to explain the increased MD and decreased FA values seen in the thoracic spine of patients with definite HAM. ⁷ , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ Our results indicate that both methods (thoracic index and DTI derivative measures) evaluated herein may be applicable in the diagnosis of definite HAM; however, additional studies involving age‐matched patients and a larger control group are essential to enhancing the diagnostic value of these parameters. Nonetheless, the findings presented herein clearly show that DTI offers high sensitivity in the detection of spinal cord damage and should be considered as a relevant diagnostic indicator in the evaluation of HAM. Moreover, correlations between DTI derivative measures and OMDS and EDSS scale results further corroborate the potential of FA and MD as biomarkers of the severity of spinal cord injury in HTLV‐1‐infected subjects.

Author Contributions

S.N.F., J.A.C.N., J.O.F., and E.M.C. contributed to conception and design of the study and methods. S.N.F., G.L.R., C.J.V.O., C.A.N., L.P., J.O.F., and E.M.C. contributed to acquisition and analysis of data. S.N.F., G.L.R., J.O.F., and E.M.C. contributed to drafting the text and preparing the figures.

Conflict of Interest

Nothing to report.

Funding Statement

This work was funded by Brazilian National Research Council (CNPq); Fundação de Amparo Pesquisa do Estado da Bahia (FAPESB); Maria Emília Pedreira Freire de Carvalho Foundation; Fundação de Amparo à Pesquisa do Estado da Bahia ; National Research Council .