Magnetic Resonance Neurography VisualizesAbnormalities in Sciatic and Tibial Nerves in PatientsWith Type 1 Diabetes and NeuropathyMichael Vaeggemose,1,2 Mirko Pham,3 Steffen Ringgaard,4 Hatice Tankisi,5 Niels Ejskjaer,6 Sabine Heiland,7

Per L. Poulsen,8 and Henning Andersen1,9

Diabetes 2017;66:1779–1788 | https://doi.org/10.2337/db16-1049

This study evaluates whether diffusion tensor imagingmagnetic resonance neurography (DTI-MRN), T2 relaxa-tion time, and proton spin density can detect and gradeneuropathic abnormalities in patients with type 1 diabetes.Patients with type 1 diabetes (n = 49) were included—11with severe polyneuropathy (sDPN), 13 with mild poly-neuropathy (mDPN), and 25 without polyneuropathy(nDPN)—along with 30 healthy control subjects (HCs).Clinical examinations, nerve conduction studies, and vi-bratory perception thresholds determined the presenceand severity of DPN. DTI-MRN covered proximal (sciaticnerve) and distal (tibial nerve) nerve segments of the lowerextremity. Fractional anisotropy (FA) and the apparentdiffusion coefficient (ADC) were calculated, as were T2relaxation time and proton spin density obtained fromDTI-MRN. All magnetic resonance findings were relatedto the presence and severity of neuropathy. FA of thesciatic and tibial nerves was lowest in the sDPN group.Corresponding with this, proximal and distal ADCs werehighest in patients with sDPN compared with patientswith mDPN and nDPN, as well as the HCs. DTI-MRNcorrelated closely with the severity of neuropathy, dem-onstrating strong associations with sciatic and tibial nervefindings. Quantitative group differences in proton spindensity were also significant, but less pronounced thanthose for DTI-MRN. In conclusion, DTI-MRN enables de-tection in peripheral nerves of abnormalities related toDPN, more so than proton spin density or T2 relaxation

time. These abnormalities are likely to reflect pathology insciatic and tibial nerve fibers.

Diabetic peripheral neuropathy (DPN) is a common com-plication that often remains undiagnosed until later stages.DPN causes irreversible damage to the peripheral nerves.Thus, early diagnosis is important in order to prevent pro-gression of DPN, emphasizing the need for more sensitivediagnostic techniques.

A DPN diagnosis is established based on a neurolog-ical examination, nerve conduction studies (NCSs), andquantitative sensory testing. DPN is associated withstructural changes of the peripheral nerves, includingendoneurial microangiopathy (1), abnormal Schwanncells (2), axonal degeneration (3), and paranodal demye-lination, all conditions leading to loss of myelinated andunmyelinated fibers (4).

Peripheral nerve lesions can be visualized by ultrasonog-raphy (5–10) and magnetic resonance neurography (MRN)(11–15). MRN enables microstructural imaging of periph-eral nerves at the anatomical level of the nerve fascicles.High-field clinical scanners (3-T) and proton spin density(PD) or T2-weighted imaging sequences with fat suppres-sion have shown an increased magnetic resonance (MR)signal in a variety of focal and nonfocal neuropathies andpolyneuropathies (11–13,16).

1Department of Neurology, Aarhus University Hospital, Aarhus, Denmark2Danish Diabetes Academy, Odense, Denmark3Department of Neuroradiology, Würzburg University Hospital, Würzburg, Germany4MR Research Centre, Aarhus University Hospital, Aarhus, Denmark5Department of Clinical Neurophysiology, Aarhus University Hospital, Aarhus, Denmark6Departments of Clinical Medicine and Endocrinology, Aalborg University Hospital,Aalborg, Denmark7Department of Neuroradiology, Heidelberg University Hospital, Heidelberg, Germany8Department of Endocrinology and Internal Medicine, Aarhus University Hospital,Aarhus, Denmark

9International Diabetic Neuropathy Consortium (IDNC), Aarhus University,Aarhus, Denmark

Corresponding author: Michael Vaeggemose, mivaeg@clin.au.dk.

Received 29 August 2016 and accepted 17 April 2017.

Clinical trial reg. no. NCT01847937, clinicaltrials.gov.

© 2017 by the American Diabetes Association. Readers may use this article aslong as the work is properly cited, the use is educational and not for profit, and thework is not altered. More information is available at http://www.diabetesjournals.org/content/license.

Diabetes Volume 66, July 2017 1779

TECHNOLOGIC

ALADVANCES

https://doi.org/10.2337/db16-1049http://crossmark.crossref.org/dialog/?doi=10.2337/db16-1049&domain=pdf&date_stamp=2017-06-03mailto:mivaeg@clin.au.dkhttp://www.diabetesjournals.org/content/licensehttp://www.diabetesjournals.org/content/license

Inclusion of diffusion tensor imaging (DTI) may improvethe diagnostic accuracy of MRN in the ulnar and mediannerves (17,18). Furthermore, in a previous study of patientswith severe DPN (sDPN), we found that DTI is a highly re-producible method that can detect sDPN in patients with type1 diabetes (19). In the same pilot study, we found that thequantitative target measures of DTI-MRN more accuratelyseparated groups than did quantitatively evaluating changesbased on T2 or PD contrast. This suggests that DTI may bethe most sensitive noninvasive imaging method to detectmicrostructural alterations of peripheral nerves in DPN.

The aim of the current study was to evaluate whetherMRN of the sciatic and tibial nerves can be used to detectand stage DPN in patients with type 1 diabetes. Further-more, applying receiver operating characteristic (ROC) ana-lyses, we aimed to determine the sensitivity and specificityof the MR methodology.

RESEARCH DESIGN AND METHODS

This study was approved by the local ethics committee (no.37251) and registered at www.clinicaltrials.gov (identifierNCT01847937). All study participants gave informedconsent.

Study PopulationThrough public announcements we recruited 49 patientswith type 1 diabetes from the Department of Endocrinologyand Internal Medicine (Aarhus University Hospital) and30 healthy control subjects (HCs). All participants wereexamined between September 2013 and May 2016. Findingsfrom 21 of these patients and 10 HCs were reported in aprevious feasibility study evaluating whether DTI enableddetection of DPN (19). That study included only patientswith sDPN or with no neuropathy (nDPN). We now extendthese findings by evaluating a larger cohort includingpatients with milder DPN. All subjects included were aged60–71 years.

The minimal criteria (20,21) determined the presence ofdiabetic neuropathy. Presence of neuropathy is determinedby at least two abnormal findings in the following fourcategories (one of these categories had to be abnormal vi-bratory perception threshold [VPT] or nerve conductionstudy [NCS]): 1) abnormal VPT ($98th percentile) at theindex finger and great toe; 2) abnormal nerve recordings(more than two) from NCSs; 3) a neuropathy symptomscore (NSS) $1; and 4) a neurological impairment score(NIS) $7. Patients with diabetes were subsequently di-vided into three groups: nDPN, mild neuropathy (mDPN)(NIS ,24), and sDPN (NIS $24).

Exclusion criteria were severe cardiac or lung disease,acute or chronic musculoskeletal disorders, acute metabolicdysregulation, other neurological or endocrine disorders,any previous or current asymmetric, proximal lower-limbweakness, and contraindications to MRI. HCs completedthe Michigan Neuropathy Screening Instrument; thisallowed us to exclude subjects with any symptoms ofneuropathy and/or diabetes (22).

Blood SamplesBlood samples were collected to measure HbA1c (milli-moles per mole) using standard laboratory methods.Plasma blood glucose was measured using a handheld glucosemeter (FreeStyle Lite; Abbot Diabetes Care, Copenhagen,Denmark).

Clinical ExaminationsClinical examinations were performed by a trained neurol-ogist (H.A.) using the NIS (23) and the NSS. The NIS is acombined score obtained from a neurological examinationof muscle strength, tendon reflex activity, and sensation atthe great toe and index finger. The NSS evaluates motor,sensory, and autonomic symptoms of neuropathy.

VPTVPT were measured at the distal part of the hallux andindex finger using the 4-2-1 stepping algorithm (CASE IV;WR Medical Electronics Co., Stillwater, MN) (24,25).

NCSsNCSs were performed with conventional surface electrodetechniques using electromyography equipment (Dantec Key-point Focus EMG version 2.11; Natus Medical, San Carlos,CA); skin temperature was $32°C during these studies. Re-sults were compared with those from laboratory controls.Motor NCSs were performed in the median, peroneal, andtibial nerves, and nerve conduction velocities (NCVs) andcompound muscle action potential (CMAP) amplitudeswere determined. Sensory NCV and sensory nerve actionpotential amplitudes were determined in the median andsural nerves. The peroneal, tibial, and sural nerves wereexamined bilaterally. The motor and sensory ulnar nerveswere studied in persons with carpal tunnel syndrome.

Neuropathy Rank Sum ScoreTo define the severity of neuropathy in each patient, wecalculated a neuropathy rank sum score (NRSS) based onthe individual rank scores from the NIS, NSS, VPT mea-surements, and NCSs.

MRNMR examinations were performed using a 3-T MR scanner(Skyra; Siemens AG, Erlangen, Germany) with a 15-channeltransmit/receive knee coil (Siemens AG). MRIs were ac-quired at predetermined locations along the left leg, in-cluding the distal thigh (10% of the distance from the upperpart of the patella to the trochanter major) and the midcalf(50% of the distance from the lateral malleolus to the lowerpoint of the patella). At the proximal level (i.e., distal thigh),the entire sciatic nerve cross section was included as aregion of interest. This location is referred to throughoutthis article as “sciatic,” whereas the level at the calf wherethe region of interest comprised only the fascicles of thetibial nerve is referred to as “tibial.”

The MR protocol was performed unilaterally and con-sisted of spin-echo (SE) images with 10 different echotimes and diffusion-weighted images to calculate diffusion

1780 DTI to Detect Neuropathy in Type 1 Diabetes Diabetes Volume 66, July 2017

http://www.clinicaltrials.gov

parameters (mean apparent diffusion coefficient [ADC],fractional anisotropy [FA], and trace). ADC and the FAimages were analyzed with nerve segmentation from thetrace images (Fig. 1). MRIs were acquired from the follow-ing pulse sequences:

1. Axial, multi-SE, 2-dimensional spectral adiabatic inver-sion recovery sequence with a strong fat suppressionpulse: repetition time (TR) = 3,280 ms; echo time(TE)1–10 = 13, 25, 38, 51, 63, 76, 89, 101, 114, and127 ms; field of view 160 3 160 mm2; matrix size512 3 512; slice thickness 3 mm; voxel size 0.33 0.333 mm3; no interslice gap, 2 averages, and 16 slices; scantime 22 min 10 s

2. Axial diffusion-weighted, SE, 2-dimensional echo-planarimaging sequence with a strong fat suppression pulse:TR = 4,200 ms; TE = 112 ms; b = 0 and 800 s/mm2;directions = 12; field of view 175 3 175 mm2; matrixsize 128 3 128; slice thickness 3 mm; voxel size 1.36 31.363 3mm3; no interslice gap, 4 averages, and 16 slices;scan time 3 min 30 s

The net imaging time was 26 min 10 s at each location,with the inclusion of an anatomical localizer scan (30 s) at eachlocation, and a coil reposition requiring an additional 5 min.

Structural fiber loss is conceived as the dominant his-tological alteration of DPN and is typically most severedistally (26); it has frequently been evaluated in sural nerve

Figure 1—Axial MRIs (TR = 3,280 ms; TE = 63 ms) (rows 1 and 2) and DTI trace images (rows 3 and 4) of the thigh (rows 1 and 3) and the shin(rows 2 and 4), demonstrating the tibial (T) and peroneal (P) nerves of an HC (A), a patient with diabetes but nDPN (B), and a patient with diabetesand sDPN (C). In A.1, a small blood vessel (V) is seen close to the nerves. A color map has been applied to increase visual contrast in the DTItrace images. The MRIs indicate enlargement and signal hyperintensity in the nerves of patients with sDPN (C); the difference is most pro-nounced for the sciatic nerve (C.1 and C.3).

diabetes.diabetesjournals.org Vaeggemose and Associates 1781

biopsies obtained at the ankle (27). To test a possible lon-gitudinal gradient of fiber loss, we evaluated the ratio of thesciatic (proximal) level versus the tibial (distal) level.

The cross-sectional area (CSA) of the segmented nervesin the multi-SE images was used to evaluate the nervecaliber of the sciatic and tibial nerves. The multi-SE imageshad a higher resolution and signal-to-noise ratio than imagesfrom DTI, providing improved structural visualization.

Imaging Processing and SegmentationFSL was used to process and analyze images (FMRIBSoftware Library, Oxford, U.K.) (28). Nerve lesions weredetermined based on quantitative analyses of signal inten-sities of the sciatic and tibial nerves. Nerves in the MRIswere segmented using the median TE (63 ms), providing asegmentation mask for the remaining multi-SE images. T2relaxation-time (T2) and PD were calculated using a mono-exponential curve-fitting algorithm applied to the nervesignal intensities of the multi-SE images (Eq. 1):

SðTEÞ ¼ ks0$exp�2

TET2

�; (1)

where k is the signal gain produced from the scanner and s0is the true PD. However, because k is difficult to separate

from the true PD, k and s0 are combined and used as PD inthis study. Monoexponential parameters were calculatedfrom the 10 different TE values acquired from the multi-SEsequence using MATLAB 2014a (MathWorks Inc.).

StatisticsThe Student t test was applied for pairwise comparisonbetween single groups, and one-way ANOVA was appliedto determine statistical differences between groups. Statis-tical significance was defined as a two-tailed P value ,0.05.Linear regression analyses were performed to evaluate anassociation between the degree of neuropathy and the DTIparameters. The goodness of fit of the linear approximationwas determined using the coefficient of determination, de-scribed as the R2 value. ROC analyses and area under thecurve (AUC) of the FA and ADC values were calculated andthen categorized according to the following predefined AUCthresholds: 1.0–0.90, excellent; 0.90–0.80, good; 0.80–0.70,fair; 0.70–0.60, poor; and 0.60–0.50, fail (29). Statisticalanalyses were performed using STATA 13.1 (StataCorpLP, College Station, TX).

RESULTS

Clinical Examinations and DemographicsThe clinical and demographic results are presented in Table 1.

Table 1—Clinical data from NCSs and quantitative sensory examinations

HCs(n = 30)

Patients with type 1 diabetes

nDPN(n = 25)

mDPN(n = 13)

sDPN(n = 11)

Age (years) 64.3 6 3.2 64.6 6 3.6 64.7 6 2.4 66.9 6 3.5

Male, n (%) 15 (50) 11 (44) 7 (54) 9 (82)

BMI (kg/m2) 26.3 6 4.4 25.0 6 3.4 27.2 6 4.2 28.0 6 4.4

Diabetes duration (years)* — 30.0 6 12 39.2 6 15.0 43.7 6 14.2

p-Glucose (mmol/L)* — 8.9 6 3.4 11 6 2.6 12 6 3.4

HbA1c (%) — 7.4 6 0.9 7.8 6 0.8 7.8 6 0.7

HbA1c (mmol/mol) — 57.2 6 10.0 61.8 6 8.4 62.0 6 7.9

NIS** — 6.4 6 6.9 15.8 6 5.3 32.7 6 6.9

NSS** — 0.4 6 0.6 1.4 6 1.1 2.8 6 2.3

VPTHand (percentile) — 81.7 6 24.5 93.1 6 7.5 93.3 6 9.3Foot (percentile) — 74.0 6 20.4 76.6 6 17.5 70.3 6 26.9

NCSsSural NCV** — 21.1 6 0.8 22.4 6 0.8 22.4 6 1.3Sural SNAP** — 20.5 6 1.5 23.1 6 2.2 23.8 6 2.3Median NCV (sensory) — 22.5 6 2.0 22.8 6 1.1 24.0 6 2.2Median SNAP — 22.7 6 2.5 23.3 6 2.3 24.1 6 3.1Tibial NCV** — 21.4 6 1.0 22.2 6 1.4 22.9 6 1.3Tibial CMAP** — 20.1 6 1.0 22.9 6 4.1 24.0 6 4.7Peroneal NCV** — 21.3 6 1.1 23.7 6 3.8 23.4 6 2.0Peroneal CMAP** — 20.1 6 0.9 22.6 6 3.0 23.4 6 2.8Median NCV (motor)* — 21.8 6 1.1 22.8 6 0.8 22.2 6 1.5Median CMAP** — 20.4 6 0.9 20.3 6 1.2 21.8 6 2.0NCS mean** — 21.1 6 0.7 22.6 6 1.2 23.2 6 1.8

Values are mean6 SD unless otherwise indicated. Results from the NCSs are given as SDs of the mean of matched HCs (age, sex, height).Values from patient groups were tested against the patients with type 1 diabetes without neuropathy. p-Glucose, plasma glucose. SNAP,sensory nerve action potential. *P , 0.05; **P , 0.01.

1782 DTI to Detect Neuropathy in Type 1 Diabetes Diabetes Volume 66, July 2017

DTI–MRN

DTIPairwise comparisons were calculated for DTI parameters ofthe sciatic and tibial nerves (Fig. 2). Significant differenceswere found between groups, indicating that FA values de-crease and ADC values increase according to severity ofneuropathy in the groups (P , 0.01). No difference wasobserved when comparing patients with no neuropathywith HCs.

Sensitivity and SpecificityTo determine the AUC, sensitivity, and specificity ofDTI in order to separate diabetic groups, ROC curveswere calculated in three analyses:

1. Patients with mDPN compared with patients with nDPN2. Patients with sDPN compared with patients with mDPN3. Patients with sDPN compared with patients with nDPN

ROC curves were calculated for FA and ADC values at thesciatic and tibial nerves (Fig. 3). On the basis of the FAvalues, there was a good separation between groups forthe sciatic nerve (AUC 0.60–0.95) and the tibial nerve

(AUC 0.69–0.90). Corresponding to this, for ADC valuesthere was also a good separation between the groups forthe sciatic nerve (AUC 0.63–0.70) and the tibial nerve (AUC0.59–0.78).

Associations Between VariablesClose correlations could be established between DTI pa-rameters and severity of neuropathy (NRSS) in the tibialand sciatic nerves (FA: R2 = 0.32 and 0.49, respectively;ADC: R2 = 0.15 and 0.19, respectively) (Fig. 4).

DTI parameters were related to the amplitude of theCMAP of the sciatic and tibial nerves (FA: R2 = 0.17 and0.24, respectively; ADC: R2 = 0.27 and 0.04, respectively)and the NCV of the sciatic and tibial nerves (FA: R2 = 0.18and 0.37, respectively; ADC: R2 = 0.31 and 0.22, respec-tively) (Table 2). Furthermore, we evaluated the relationsbetween DTI parameters and the NIS (FA: R2 = 0.33 and0.28; ADC: R2 = 0.14 and 0.07) (Table 2).

Distal-to-Proximal GradientThe ratio between DTI findings from the distal (tibial) nerveand the proximal (sciatic) nerve was similar between thegroups, as indicated by the FA ratio (HCs: 0.87 6 0.12;

Figure 2—Box plots of FA (top) and ADC (bottom) comparing HCs and patients with type 1 diabetes with no neuropathy (No DPN), with mildneuropathy (Mild DPN), and with severe neuropathy (Severe DPN). FA values have no unit; ADC values are 1023 mm2/s. P values representstatistical differences from pairwise comparisons (Student t test). The plots illustrate the 25th and 75th percentiles (boxes), adjacent values(whiskers), outliers (dots), and median values of the groups (black horizontal lines in grey boxes).

diabetes.diabetesjournals.org Vaeggemose and Associates 1783

nDPN group: 0.87 6 0.14; mDPN group: 0.85 6 0.15;sDPN group: 0.81 6 0.16) (P = 0.50) and the ADCratio (HCs: 1.04 6 0.15; nDPN group: 1.04 6 0.12;mDPN group: 1.08 6 0.24; sDPN group: 1.11 6 0.14)(P = 0.67).

However, a statistically significant difference was foundin the distal-to-proximal gradient of the FA values betweenthe three groups; this was found only for the ADC values inthe sDPN group (P = 0.05) (Table 3).

MR Signal Analyses

T2 and PDT2 and PD of the sciatic and tibial nerves showed nodifferences between groups (Table 4). In pairwise analysesof the PD, we found a difference between the nDPN andsDPN groups for the sciatic nerve (P = 0.03); a similardifference was found between these same groups for thetibial nerve (P = 0.03).

Nerve CaliberCSAs in the sciatic and tibial nerves were different betweengroups (Table 4). Paired sample t tests revealed that theHCs had smaller sciatic nerve CSAs compared with allgroups of patients with diabetes (P = 0.01). For the tibialnerve, the CSA was different when comparing the nDPNand mDPN groups (P = 0.01) and for the HCs comparedwith the mDPN group (P = 0.04).

DISCUSSION

We established that MRN is able to detect in patients withtype 1 diabetes nerve abnormalities that are closely relatedto the severity of neuropathy, suggesting that MRN can beused to detect structural signs of neuropathy. In this studywe extended previous studies investigating MRN in DPN(11,12) by exclusively evaluating DPN in patients with type1 diabetes and, importantly, by incorporating advancedMRN with DTI, to our knowledge for the first time. Ourmain finding is the superior diagnostic accuracy of quanti-tative DTI over PD or T2 values. In line with previousstudies, we found that the structural nerve differences asvisualized by imaging appear most marked at the proximalrather than the distal level.

In the Toronto criteria, NCS remains the gold standardfor diagnosing and grading DPN in clinical research, and newdiagnostic methods should be compared with and relatedto findings of NCSs (30). We established that MRNand NCSs were closely associated, suggesting that MRNreflects the pathophysiological process of DPN in type 1diabetes. In a previous study we established that the DTI-MRN techniques applied are highly reproducible and reli-able (19).

In this study, MRI included DTI (FA and ADC) andmulti-SE imaging. FA and ADC describe the restriction ofwater molecule diffusion in three dimensions. FA reflects

Figure 3—ROC curves of the sensitivity and specificity of FA (top) and ADC (bottom) values in the sciatic and tibial nerves.

1784 DTI to Detect Neuropathy in Type 1 Diabetes Diabetes Volume 66, July 2017

restrictions of spatial movement and ADC reflects the diffusionspeed of the water molecules. In peripheral neuropathies, loss ofaxons and myelin leads to less constriction of endoneurial flowalong the nerves (31); this could explain the changes in FAand ADC. Experimental studies have shown that FA corre-lates to nerve fiber density (32–36). ADC changes in rela-tion to membrane, myelin sheath, cell wall, macromolecule,and viscosity alterations in the fluid containing low levels ofprotein flowing along the nerve fibers (31). In DPN, axonalloss is considered the most prevalent pathological finding(3), suggesting that FA is a good measure of DPN.

The presence and severity of neuropathy were deter-mined from NCSs, quantitative sensory examinations, andclinical examinations. Pairwise comparisons between thegroups with diabetes and the HCs showed that FA and ADCvalues of the sciatic and tibial nerves had the highestdiscriminatory power between nDPN, subtle DPN, andsDPN. The FA values differed significantly between groups,and even mDPN could be separated from nDPN. Interest-ingly, this difference was more pronounced at the proximal/sciatic level than at the distal level, which is also reflectedby the AUC of the ROC analyses. The proximal domi-nance of structural alteration in this study is consistentwith previous studies showing predominantly proximalstructural nerve injury not only in DPN but also in otherpolyneuropathies with similar distal symmetric symptoms(12,13). Patients with diabetes without neuropathy hadADC and FA values similar to those in HCs, indicatingthat the abnormal FA and ADC values reflect neuropathic

abnormalities rather than an effect of diabetes per se. Onthe basis of the AUC from the FA and ADC values obtainedat the proximal/sciatic level (Fig. 3), FA had the best dis-criminatory performance compared with ADC.

Linear regression analyses demonstrated close associa-tions between the severity of neuropathy and the FA valuesof the sciatic and tibial nerves. The association was lesspronounced in relation to the ADC values.

Furthermore, FA and ADC of the sciatic and tibial nervesshowed good associations with NCV and CMAP, with theclosest correlations for the FA values of the sciatic nerve.Evaluating the relation between DTI and NIS, FA was moreclosely related than ADC.

Figure 4—Linear regression analyses of FA (top) and ADC (bottom) values and their associations with the NRSS in the sciatic (left) and tibial(right) nerves. DM1, type 1 diabetes.

Table 2—Associations between DTI parameters (FA and ADC)and NCSs and clinical examinations

Sciatic nerve Tibial nerve

R2 P value R2 P value

FA vs. CMAP 0.24 ,0.01 0.17 ,0.01

FA vs. NCV 0.37 ,0.01 0.18 0.01

FA vs. NIS 0.28 ,0.01 0.33 ,0.01

ADC vs. CMAP 0.04 0.13 0.27 ,0.01

ADC vs. NCV 0.22 0.01 0.31 ,0.01

ADC vs. NIS 0.07 0.08 0.14 0.01

NCSs consist of determining the CMAP and NCV of the peronealand tibial nerves. Linear regression analyses were applied forcorrelation analyses.

diabetes.diabetesjournals.org Vaeggemose and Associates 1785

DTI parameters have been used to evaluate axonal andmyelin sheath integrity in the median nerve in healthysubjects (37). These results were validated in part by in-dicating that FA and ADC correlated with the integrity ofthe myelin sheath identifiers (NCV) and axial diffusivitywith fiber density/axonal integrity (CMAP) (18,38). In an-other study of patients with both axonal and demyelinatingneuropathies, the FA of the sciatic nerve was closely relatedto CMAP (39). Thus, in this and in previous studies, FAenabled more sensitive and accurate detection of peripheralnerve abnormalities than ADC (40), irrespective of neuro-pathic severity.

Distal-to-proximal loss of axons was evaluated betweenand within groups based on the DTI parameters. We founda within-group difference in the proximal and distal FAvalues, in line with earlier findings (37). This underlines theimportance of using predefined anatomical scanning locations.

In our study, the CSA of the sciatic nerve was lower inHCs compared with patients with diabetes. For the tibialnerve, patients with mDPN had a larger CSA than patientswith nDPN; however, patients with sDPN did not have alarger CSA. This is in contrast to findings using ultraso-nography, where enlarged nerves occurred distally inadvanced stages of neuropathy (5). The resolution usedfor segmentation of the CSA was 0.3 3 0.3 mm2/pixel.

Because the average CSA of the tibial nerve is 7 mm2, .75pixels are used for segmentation (225 pixels in the sciaticnerve). This suggests that the different findings using ul-trasound and MRI cannot be explained by a lower resolu-tion of MRI. Our finding is consistent with previous studiesshowing that an increase in CSA occurs predominantly atthe proximal level, in addition to DTI and signal alter-ations (12,13). Ultrasonography is excellent for high-reso-lution imaging of superficial peripheral nerves; however, itdoes not enable visualization of deeply situated nerves,nerves surrounded by fat (sciatic nerve), or nerves beneathbones (because of acoustic artifacts) (41,42).

Previous studies have shown that MRN is a more reliablemeasure than ultrasonography for use to visualize lesions(43). Nevertheless, because MRN is more expensive and timeconsuming, ultrasonography may be a more feasible method.

To detect DPN, ultrasonography is commonly used todetermine CSA; however, hypoechogenicity and maximumthickness of nerve fascicles have also been evaluated (6,8).These studies indicated a correlation between NCSs andultrasound findings. However, other studies have notbeen able to find a correlation between CSA and find-ings from NCSs (7,44). Ultrasonography may enable earlydetection of subclinical nerve degeneration, but because ofthe higher resolution, MRN may also be relevant.

Table 3—Distal-to-proximal gradients of FA and ADC values in HCs and patients with type 1 diabetes with nDPN, mDPN, and sDPN

Sciatic nerve Tibial nerve Difference Difference 95% CI P value

FAHC 0.48 6 0.06 0.42 6 0.06 0.06 6 0.06 0.04–0.09 ,0.01nDPN 0.47 6 0.04 0.41 6 0.07 0.06 6 0.06 0.03–0.09 ,0.01mDPN 0.41 6 0.07 0.34 6 0.06 0.06 6 0.06 0.03–0.10 ,0.01sDPN 0.38 6 0.04 0.31 6 0.08 0.08 6 0.06 0.03–0.12 ,0.01

ADCHC 1.47 6 0.16 1.52 6 0.19 20.05 6 0.22 20.14 to 0.03 0.22nDPN 1.52 6 0.12 1.59 6 0.19 20.06 6 0.19 20.14 to 0.02 0.12mDPN 1.63 6 0.26 1.74 6 0.42 20.11 6 0.39 20.35 to 0.12 0.31sDPN 1.62 6 0.17 1.78 6 0.20 20.16 6 0.22 20.32 to 0.00 0.05

Values are mean 6 SD unless otherwise indicated. FA values have no units; ADC values are 1023 mm2/s. Paired sample t test was used toacquire P values.

Table 4—T2 relaxation time, PD, and CSA of the sciatic and tibial nerves in HCs and patients with type 1 diabetes with nDPN, mDPN,and sDPN

HCs

Patients with type 1 diabetes

P valuenDPN mDPN sDPN

T2 relaxation time (ms)Sciatic 79 6 8 83 6 9 82 6 16 83 6 7 0.58Tibial 61 6 10 62 6 9 63 6 13 64 6 6 0.80

PDSciatic 381 6 80 403 6 73 413 6 124 343 6 77 0.17Tibial 545 6 112 570 6 115 499 6 149 484 6 87 0.14

CSA (mm2)Sciatic 21 6 6 27 6 8 26 6 5 28 6 8 0.02Tibial 7 6 2 6 6 2 9 6 4 8 6 3 0.04

ANOVA group analyses were used to acquire P values. Values are mean 6 SD. PD values have no units.

1786 DTI to Detect Neuropathy in Type 1 Diabetes Diabetes Volume 66, July 2017

In line with previous studies, the T2 value had lowdiscriminatory power. Interestingly, PD was inferior to DTIfor group discrimination. The relation between abnormali-ties observed on DTI-MRN and the pathological processof neuropathy remains unclear. Hyperglycemia leads to cel-lular nerve damage through mitochondrial overload, polyolpathway–induced oxidative stress, and inflammatory injury(45,46). T2-weighted images are sensitive to edema and fat.We applied a strong fat saturation pulse to remove theepineural fat signal adjacent to the nerve fascicles, causingfat and connective tissue to appear dark in the MRIs.In patients with diabetes, severely damaged nerve tissueis replaced by connective tissue (47) and therefore appearsdark in SE images. This could explain the absence of T2changes in patients with sDPN.

In a previous study using a mixed group of participantswith type 1 or type 2 diabetes, the PD signal was higher inpatients with both mDPN and sDPN (12). In that study, T2-weighted and PD-weighted imaging were applied, but with anormal fat saturation pulse over a larger area. This approach,when compared with our study, which includes only patientswith type 1 diabetes, might explain the difference in findingsof PD. Therefore, a study evaluating MRN in type 2 diabetesis necessary in order to evaluate possible differences betweenthe two types of diabetes.

Axonal loss would not result in altered signal intensity ofthe T2-weighted images (12,48). FA is closely associatedwith fiber density (32–34), which may explain why FA val-ues are more closely related to the severity of neuropathycompared with T2 and PD.

Our study has several limitations. First, this is a cross-sectional study, and thus how DTI findings develop overtime remains unknown. Second, axial and radial diffusivitieswere not calculated in our study, which in HCs providedadditional information about axonal and myelin sheath in-tegrity (37). Third, MRN coverage of the sciatic and tibialnerves consisted of 16 slices (2 3 4.80 cm), evaluating onlysmall parts of the nerve. MRN covering the entire nervewould enable detection of multifocal lesions in DPN; however,this would considerably increase examination time. Also, we didnot include the upper-limb nerve to serve as a control for thelower limb, which could have further substantiated our findings;this would, however, also increase the examination time. Finally,the study did not include an assessment of peripheral limbvascular status to evaluate any influence of multifocal ischemicneuropathy on the DTI-MRN findings.

In conclusion, we found close associations betweenDTI-MRN findings and the presence and severity of neu-ropathy in proximal and distal nerve segments of patientswith type 1 diabetes. DTI-MRN is a noninvasive, quantita-tive method that may be used to detect and monitorneuropathic processes in DPN.

Acknowledgments. The authors thank Søren Gregersen, Department ofEndocrinology and Internal Medicine, Aarhus University Hospital, for helping withpatient recruitment.

Funding. This work is funded by the UNIK partnership foundation, SiemensA/G Copenhagen, Aarhus University, the BEVICA Foundation, and the DanishDiabetes Academy, which is supported by the Novo Nordisk Foundation. M.P.(SFB 1158, Project A3) and S.H. (SFB 1118 Project, B05) were supported by theDeutsche Forschungsgemeinschaft.Duality of Interest. No conflicts of interest relevant to this article werereported.Author Contributions. M.V., M.P., N.E., S.H., and H.A. designed the study.M.V. and P.L.P. recruited the patients. M.V. and H.A. examined the patients. M.V.,S.R., H.T., and H.A. performed the research. M.V. analyzed the data and wrotethe manuscript. M.V. is the guarantor of this work and, as such, had full access to allthe data in the study and takes responsibility for the integrity of the data and theaccuracy of the data analysis.

References1. Cameron NE, Eaton SEM, Cotter MA, Tesfaye S. Vascular factors and metabolicinteractions in the pathogenesis of diabetic neuropathy. Diabetologia 2001;44:1973–19882. Malik RA, Veves A, Walker D, et al. Sural nerve fibre pathology in diabeticpatients with mild neuropathy: relationship to pain, quantitative sensory testing andperipheral nerve electrophysiology. Acta Neuropathol 2001;101:367–3743. Dyck PJ, Lais A, Karnes JL, O’Brien P, Rizza R. Fiber loss is primary andmultifocal in sural nerves in diabetic polyneuropathy. Ann Neurol 1986;19:425–4394. Fernyhough P, Roy Chowdhury SK, Schmidt RE. Mitochondrial stress and thepathogenesis of diabetic neuropathy. Expert Rev Endocrinol Metab 2010;5:39–495. Riazi S, Bril V, Perkins BA, et al. Can ultrasound of the tibial nerve detect di-abetic peripheral neuropathy? A cross-sectional study. Diabetes Care 2012;35:2575–25796. Watanabe T, Ito H, Sekine A, et al. Sonographic evaluation of the peripheralnerve in diabetic patients: the relationship between nerve conduction studies, echointensity, and cross-sectional area. J Ultrasound Med 2010;29:697–7087. Pitarokoili K, Kerasnoudis A, Behrendt V, et al. Facing the diagnostic challenge:nerve ultrasound in diabetic patients with neuropathic symptoms. Muscle Nerve2016;54:18–248. Ishibashi F, Taniguchi M, Kojima R, Kawasaki A, Kosaka A, Uetake H. Mor-phological changes of the peripheral nerves evaluated by high-resolution ultraso-nography are associated with the severity of diabetic neuropathy, but not cornealnerve fiber pathology in patients with type 2 diabetes. J Diabetes Investig 2015;6:334–3429. Ishibashi F, Taniguchi M, Kojima R, Kawasaki A, Kosaka A, Uetake H. Elasticityof the tibial nerve assessed by sonoelastography was reduced before the develop-ment of neuropathy and further deterioration associated with the severity of neu-ropathy in patients with type 2 diabetes. J Diabetes Investig 2016;7:404–41210. Breiner A, Qrimli M, Ebadi H, et al. Peripheral nerve high-resolution ultrasoundin diabetes. Muscle Nerve 2017;55:171–17811. Pham M, Oikonomou D, Bäumer P, et al. Proximal neuropathic lesions in distalsymmetric diabetic polyneuropathy: findings of high-resolution magnetic resonanceneurography. Diabetes Care 2011;34:721–72312. Pham M, Oikonomou D, Hornung B, et al. Magnetic resonance neurographydetects diabetic neuropathy early and with Proximal Predominance. Ann Neurol2015;78:939–94813. Kollmer J, Hund E, Hornung B, et al. In vivo detection of nerve injury in familialamyloid polyneuropathy by magnetic resonance neurography. Brain 2015;138:549–56214. Pham M, Bäumer P, Meinck HM, et al. Anterior interosseous nerve syndrome:fascicular motor lesions of median nerve trunk. Neurology 2014;82:598–60615. Kim B, Srinivasan A, Sabb B, Feldman EL, Pop-Busui R. Diffusion tensor im-aging of the sural nerve in normal controls. Clin Imaging 2014;38:648–65416. Chhabra A, Andreisek G, Soldatos T, et al. MR neurography: past, present, andfuture. AJR Am J Roentgenol 2011;197:583–59117. Breckwoldt MO, Stock C, Xia A, et al. Diffusion tensor imaging adds diagnosticaccuracy in magnetic resonance neurography. Invest Radiol 2015;50:498–504

diabetes.diabetesjournals.org Vaeggemose and Associates 1787

18. Bäumer P, Pham M, Ruetters M, et al. Peripheral neuropathy: detection withdiffusion-tensor imaging. Radiology 2014;273:185–19319. Vaeggemose M, Pham M, Ringgaard S, et al. Diffusion tensor imaging MRneurography for the detection of polyneuropathy in type 1 diabetes. J Magn ResonImaging 2017;45:1125–113420. Dyck PJ, Kratz KM, Karnes JL, et al. The prevalence by staged severity ofvarious types of diabetic neuropathy, retinopathy, and nephropathy in a population-based cohort: the Rochester Diabetic Neuropathy Study. Neurology 1993;43:817–82421. Dyck PJ, Albers JW, Andersen H, et al.; Toronto Expert Panel on DiabeticNeuropathy. Diabetic polyneuropathies: update on research definition, diagnosticcriteria and estimation of severity. Diabetes Metab Res Rev 2011;27:620–62822. Herman WH, Pop-Busui R, Braffett BH, et al.; DCCT/EDIC Research Group. Useof the Michigan Neuropathy Screening Instrument as a measure of distal symmet-rical peripheral neuropathy in type 1 diabetes: results from the Diabetes Control andComplications Trial/Epidemiology of Diabetes Interventions and Complications. DiabetMed 2012;29:937–94423. Dyck PJ, Hughes RAC, O’Brien PC. Quantitating overall neuropathic symptoms,impairments, and outcomes. In Peripheral Neuropathy. Vol. 1. Elsevier Inc., 2005, p.1031–105124. Dyck PJ, Litchy WJ, Lehman KA, Hokanson JL, Low PA, O’Brien PC. Variablesinfluencing neuropathic endpoints: the Rochester Diabetic Neuropathy Study ofHealthy Subjects. Neurology 1995;45:1115–112125. Dyck PJ, O’Brien PC, Kosanke JL, Gillen DA, Karnes JL. A 4, 2, and 1 steppingalgorithm for quick and accurate estimation of cutaneous sensation threshold.Neurology 1993;43:1508–151226. Said G. Diabetic neuropathy. Handb Clin Neurol 2013;115:579–58927. Malik RA. The pathology of human diabetic neuropathy. Diabetes 1997;46(Suppl. 2):S50–S5328. Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM. FSL.Neuroimage 2012;62:782–79029. Xia J, Broadhurst DI, Wilson M, Wishart DS. Translational biomarker discoveryin clinical metabolomics: an introductory tutorial. Metabolomics 2013;9:280–29930. Tesfaye S, Boulton AJM, Dyck PJ, et al.; Toronto Diabetic Neuropathy ExpertGroup. Diabetic neuropathies: update on definitions, diagnostic criteria, estimation ofseverity, and treatments. Diabetes Care 2010;33:2285–229331. Filler AG, Maravilla KR, Tsuruda JS. MR neurography and muscle MR imagingfor image diagnosis of disorders affecting the peripheral nerves and musculature.Neurol Clin 2004;22:643–682, vi–vii32. Takagi T, Nakamura M, Yamada M, et al. Visualization of peripheral nervedegeneration and regeneration: monitoring with diffusion tensor tractography. Neu-roimage 2009;44:884–892

33. Lehmann HC, Zhang J, Mori S, Sheikh KA. Diffusion tensor imaging to assessaxonal regeneration in peripheral nerves. Exp Neurol 2010;223:238–24434. Wan Q, Wang S, Zhou J, et al. Evaluation of radiation-induced peripheral nerveinjury in rabbits with MR neurography using diffusion tensor imaging and T2measurements: correlation with histological and functional changes. J Magn ResonImaging 2016;43:1492–149935. Rosser BA, Vowles KE, Keogh E, Eccleston C, Mountain GA. Technologically-assisted behaviour change: a systematic review of studies of novel technologies forthe management of chronic illness. J Telemed Telecare 2009;15:327–33836. Beaulieu C. The basis of anisotropic water diffusion in the nervous system - atechnical review. NMR Biomed 2002;15:435–45537. Heckel A, Weiler M, Xia A, et al. Peripheral nerve diffusion tensor imaging:assessment of axon and myelin sheath integrity. PLoS One 2015;10:e013083338. Kakuda T, Fukuda H, Tanitame K, et al. Diffusion tensor imaging of peripheralnerve in patients with chronic inflammatory demyelinating polyradiculoneuropathy: afeasibility study. Neuroradiology 2011;53:955–96039. Mathys C, Aissa J, Meyer Zu Hörste G, et al. Peripheral neuropathy: assessmentof proximal nerve integrity by diffusion tensor imaging. Muscle Nerve 2013;48:889–89640. Khalil C, Budzik JF, Kermarrec E, Balbi V, Le Thuc V, Cotten A. Tractography ofperipheral nerves and skeletal muscles. Eur J Radiol 2010;76:391–39741. Goedee HS, Brekelmans GJF, van Asseldonk JTH, Beekman R, Mess WH,Visser LH. High resolution sonography in the evaluation of the peripheral nervoussystem in polyneuropathy–a review of the literature. Eur J Neurol 2013;20:1342–135142. Beekman R, Visser LH. High-resolution sonography of the peripheral nervoussystem–a review of the literature. Eur J Neurol 2004;11:305–31443. Pham M, Bäumer T, Bendszus M. Peripheral nerves and plexus: imagingby MR-neurography and high-resolution ultrasound. Curr Opin Neurol 2014;27:370–37944. Hobson-Webb LD, Massey JM, Juel VC. Nerve ultrasound in diabetic poly-neuropathy: correlation with clinical characteristics and electrodiagnostic testing.Muscle Nerve 2013;47:379–38445. Vincent AM, Russell JW, Low P, Feldman EL. Oxidative stress in the patho-genesis of diabetic neuropathy. Endocr Rev 2004;25:612–62846. Callaghan BC, Cheng HT, Stables CL, Smith AL, Feldman EL. Diabetic neu-ropathy: clinical manifestations and current treatments. Lancet Neurol 2012;11:521–53447. Muona P, Peltonen J. Connective tissue metabolism in diabetic peripheralnerves. Ann Med 1994;26:39–4348. Bendszus M, Stoll G. Technology insight: visualizing peripheral nerve injuryusing MRI. Nat Clin Pract Neurol 2005;1:45–53

1788 DTI to Detect Neuropathy in Type 1 Diabetes Diabetes Volume 66, July 2017