Total neuropathy score

Methods.

In this single-center study, subjects with diabetic polyneuropathy were drawn from the investigators’ practices. Baseline neuropathy was assessed before randomization by reviewing each subject’s symptoms with one of the investigators. Healthy male and female subjects between 18 and 75 years of age without evidence of neuropathy or with mild, moderate, or severe diabetic polyneuropathy were selected to represent the entire range of possible TNSs. All subjects gave written informed consent. The protocol was approved by the Joint Committee on Clinical Investigation of the Johns Hopkins University School of Medicine and the Johns Hopkins Hospital.

The randomization schedule was developed to balance the potential order effect of the evaluation sequence. Patients’ randomization were stratified by their screening diabetic polyneuropathy assessment: normal, mild, moderate, or severe. The evaluation sequence was then assigned randomly to each patient within a stratum according to a computer-generated randomization list balanced for rater (1 or 2) and order of procedures (TNS or NIS and NSS).

All testing was performed once at baseline and then 1 to 3 weeks thereafter, except for NC studies, for which reliability data already exists,13,14 and QSTs which were performed once at baseline.

Subjects were assessed by two raters independently at each visit. Raters did not communicate their findings to one another during the course of the study. NC studies were performed using our standard laboratory techniques using a Clark–Davis Advantage EMG machine (Toronto, Canada). Right sural sensory and peroneal motor nerves were studied with standardized techniques and fixed distances. The temperature was maintained at >32 °C. Vibratory thresholds were measured with Vibratron II (Physitemp Instruments, Inc.) at the right great toe and the index finger of the nondominant hand.

To assess properly the reliability of the TNS in quantifying the severity of neuropathy, only diabetic subjects were used to estimate the intrarater and interrater reliabilities.

The intrarater reliability of the TNS between visits was estimated based on data assessed by the same rater at both visits. An analysis of variance (ANOVA) approach was employed using a random-effect model to estimate the intrarater reliability for each rater. Overall intrarater reliability was assessed by ignoring rater distinction and pooling data from both raters. The interrater reliability of the TNS between raters was estimated based on data assessed by both raters at the same visit. An ANOVA approach was employed using a random-effect model. Per the study design, each subject was to be assessed by both raters only at one of the two study visits. The assessments collected at that particular visit were used for the estimation of the interrater reliability. Of the 30 diabetic patients with neuropathy, 15 patients were assessed by both raters at visit 1, and the other 15 were assessed at visit 2. Because both raters assessed the same patient at the same visit, data collected during visit 1 and visit 2 were considered “poolable” for the estimation of the overall interrater reliability. One-sided 95% CIs were also constructed for both the intra- and interrater reliability.15

For cross-sectional validation, Spearman’s rank correlation coefficient was used for estimating the conformity between the TNS and the NIS and NSS, due to violations of normality. The measurement scores of the TNS and the NIS and NSS were ranked in an ascending order (the higher the rank the more severe the neuropathy). Averaged ranks were used in the events of ties. Estimates of Spearman’s rank correlation coefficient were calculated for all possible correlations among the TNS and its subscales with the NIS and NSS. A 95% CI was also constructed for each correlation coefficient.

Results.

A total of 35 subjects were enrolled and randomized into the four evaluation groups balanced by rater and the etiology (diabetic or healthy) and the severity of neuropathy to ensure that the entire range of the TNS and standard measurements would be represented in each group. As a result, 7, 10, 7, and 11 patients were assigned to the four evaluation groups respectively.

Of the 35 subjects, 30 were diabetic patients with neuropathy (10 each with mild, moderate, and severe neuropathy) and 5 were normal subjects. Demographics and baseline characteristics are presented in table 2.

The overall TNS intrarater reliability was 0.973 (95% CI, ≥0.950), and the overall interrater reliability was 0.938 (95% CI, ≥0.836). Both intrarater and interrater reliability for the individual raters was high (table 3).

The TNS and NIS, and the TNS and NSS were all correlated highly and positively (table 4). In addition, all the TNS subscales were correlated highly with the NSS and NIS (see table 4). The highest correlation of the NIS occurred with examination findings, and the highest correlation of the NSS occurred with symptom evaluation, although examination findings also correlated with the NSS.

Discussion.

This study has demonstrated that the TNS is a valid measure of peripheral nerve function with high inter- and intrarater reliability, and high correlations to currently available and validated measures of neuropathy in patients with diabetes. The current TNS has been modified from the TNS that arose from our studies of chemotherapy-induced neuropathy in oncology patients.11,12 In designing a strategy for their serial neurologic evaluation, simple, inexpensive, and brief but sensitive measures of the onset and development of distal neuropathy were needed. Previously, there was no validated instrument that combined symptoms, signs, NC, and QST measures to give a composite picture of a patient’s neuropathy. The TNS provides such a composite.

Because we are interested mainly in distal length-dependent axonal neuropathy—the most prevalent form of polyneuropathy—patients are asked whether they have sensory or motor symptoms distally in the limbs, and are asked to define the severity of those symptoms. We believe that this is useful and different than the NSS and NSP. The TNS questions are specific for distally prominent neuropathies. There is no diluting effect of other, nondistal symptoms. The TNS symptom score grades the severity of the sensory, motor, or autonomic symptoms.

Like the symptom score system, the examination concentrates on addressing the presence and severity of distal length-dependent neuropathy. Dyck et al.16 and Feldman et al.2,3 have shown that distal sensation, strength, and reflexes are the most reproducible components of the NIS and the Michigan Diabetic Neuropathy Scale, and are predictive of diabetic neuropathy.17 Our method of distal neuropathy assessment is in agreement with these findings. This is also why we choose to give the examination findings the maximum weight in the TNS.

Because the emphasis of the TNS is on longitudinal assessment of distal length-dependent axonal neuropathy, we use a distal sensory (sural) and a distal motor (peroneal) amplitude measure in the TNS, and grade its severity. In most such neuropathies, latency and conduction velocity do not change much and thus do not reflect changes in the neuropathy. In addition, concentrating on these two measures alone allows us to discount unrelated NC changes (for example, those due to underlying entrapment neuropathy or radiculopathy). In fact, Dyck et al.10 have shown that sural amplitude and peroneal amplitude were correlated highly and significantly with the NSS, NDS, and, most importantly, with the neuropathology.

Lastly, the QST is a useful tool in quantifying small- and large-fiber dysfunction. Although several QST machines are now available, we use Vibratron II, an inexpensive and easy-to-use system, to assess large-fiber function.

The data from symptoms, signs, NC studies, and QSTs are each assigned a score depending on severity. Addition of these individual scores gives a TNS value that indicates the extent of neuropathy. In practice, the TNS works quite well and is a sensitive measure of peripheral nerve function. Moreover, as we have shown previously in our studies on toxic neuropathies,11,12 the TNS can be used to assess the progression of neuropathy longitudinally. We believe, therefore, that if a potential therapeutic intervention is purported to alter the progressive course of a distal dominant neuropathy, the TNS may be a useful tool to detect this.

We have chosen to validate the TNS against the NSS and NIS. Both these measures have been used frequently, including in clinical trials,18-22 and both have been validated against indices of neuropathology,10 which is considered the gold standard of peripheral nerve function. In the current study, the TNS is a valid measure of neuropathic impairment and it could be used in clinical trials in which the region of interest is the distal peripheral nerve. The TNS may not be as valuable in demyelinating neuropathies.

We have shown that the intra- and interrater reliability of the TNS is excellent. However, this was performed by two physicians (D.R.C., V.C.) who are experienced in the use of the TNS. Our results may need to be confirmed by other intra- and interrater reliability studies of larger numbers of physicians. However, we believe that any experienced neurologist can obtain TNSs reliably, because it is dependent on a standard neurologic evaluation.

We believe the TNS has important advantages over current methods used to evaluate peripheral neuropathy. The battery of studies is simple, direct, and brief. The symptom questionnaire takes approximately 3 minutes to complete, and the directed examination takes approximately 5 minutes. The NC study and the QST can be completed in approximately 30 minutes. These tests generate the primary data from which the TNS is derived. These studies are relatively inexpensive and use equipment that is readily available. The information collected concentrates on the presence and the development of distal neuropathy, which is common in many neuropathies. The TNS provides “on-line” data regarding a patient’s neuropathy, combining information from both subjective and objective test measures in a fashion not done previously. The data collected can be converted readily to a TNS value, which is available immediately. Lastly, an overall evaluation of the neuropathy is obtained, which correlates with individual measures.

Although we have validated the 10 component TNSs as shown in table 1, the TNS is potentially adaptable to suit other clinical situations in which the fiber population of interest may be different. For example, in clinical studies of painful, sensory neuropathies, select components of the TNS presented here could be replaced by others such as the symptom of pain (as opposed to worst sensory symptom), temperature sensation, or QST for cooling sensibility. The instrument used to assess QST and to quantify sensory symptoms and signs may need to be standardized as well. However, alterations such as these would require separate validation studies. In addition, with the development of new tools for assessment of peripheral neuropathy such as skin biopsy,23-25 we may be able to add another objective measure of small-fiber assessment. Because intraepidermal fiber density is quantifiable, a severity scoring system could be added to the TNS. We have also not included any objective measures of autonomic dysfunction in the TNS, and this may be necessary to aid in the assessment of neuropathies with prominent autonomic involvement.

With the validation of the TNS in diabetic neuropathy, we envision a number of future research opportunities. Although we have used the TNS successfully in monitoring and detecting the development of toxic neuropathy,11,12 the TNS has not been used in longitudinal studies of diabetic neuropathy, particularly in therapeutic studies. Additionally, determining the ability to generalize the TNS to other neuropathy populations would be useful. One of the most difficult challenges in future directions for the TNS is determination of the number of TNS points that translate into diagnosing different grades of severity of neuropathy and what constitutes a meaningful clinical change.26 We do not have data on this issue because that was not the point of this study nor our intention in developing the TNS. Because the TNS has proved useful in measuring the development of neuropathy in “healthy individuals” exposed to a potential neurotoxin, the TNS could also detect potentially the regression of neuropathy under suitable circumstances.