ORIGINAL ARTICLE

Prevalence of Detected by Current Threshold Test in Subjects with Glucose Metabolic Impairments in a Community Kiyoshi Takekuma, Fujiko Ando*, Naoakira Niino* and Hiroshi Shimokata*

Abstract Introduction Objective Recent studies reported that hyperesthesia Quantitative sensation testing methodsare recommendedfor maybe an indicator of early diabetic polyneuropathy. Us- characterizing cutaneous sensation alteration in patients with ing the current perception threshold (CPT) test, which diabetic polyneuropathy ( 1 ). These methods can detect hyper- stimulates peripheral sensory nerve fibers by three differ- esthesia (1), which is thought to be a marker of injured nerve ent frequencies (2,000, 250, and 5 Hz), we investigated the fibers (2). Recently, Dyck and coworkers reported a higher relationship between hyperesthesia and glucose metabolic prevalence of hyperesthesia in patients with a milder spectrum impairment in a community. of diabetic polyneuropathy with these methods (3), suggesting Methods The number of subjects, aged 40 to 79 years, it maybe worthwhile to assess for hyperesthesia as an indica- was 2,074. The CPTvalues at each frequency were classi- tor of diabetic polyneuropathy. fied into three categories (hyperesthesia, normal, and At present, there is little information on the natural history hypoesthesia). Subjects were also subgrouped into three of diabetic neuropathy. However, some previous studies have groups (normal, insulin resistance, and ) accord- reported the occurrence of diabetic polyneuropathy early in ing to glucose metabolic status, and those with hypoesthesia the course of diabetes (4, 5). These reports suggested the pos- at each frequency were excluded in the analyses. sibility of hyperesthesia in diabetic patients before the devel- Results The prevalence of hyperesthesia at 2,000, 250, opment of overt polyneuropathy. and 5 Hz in male diabetic subjects were 14.1, 15.6, and 7.7%, This study was aimed to examine the possibility whether respectively, and 22.2, 24.5, and 16.4% respectively in fe- hyperesthesia may develop early in the course of diabetes in male diabetic subjects. In logistic regression analysis ad- people living in a community. To clarify this issue, we investi- justed for age, females with diabetes showed a significantly gated the association between hyperesthesia and glucose meta- high odds ratio (OR) for hyperesthesia at 2,000 Hz (OR, bolic status in subjects subdivided into three groups according 2.42; 95% confidence interval (95%CI), 1.18 to 4.97) and to glucose metabolic status (normal, insulin resistance, and 250 Hz (OR, 2.65; 95%CI, 1.31 to 5.37). In male diabetic diabetes) in a community using current perception threshold subjects, a significantly high odds ratio for hyperesthesia (CPT) test, which is used to evaluate peripheral sensory nerve was seen at 250 Hz (OR, 2.09; 95%CI, 1.07 to 4.05). fibers quantitatively (6). Conclusion Ourresults suggested that hyperesthesia mayemerge coupled with developing diabetes, supporting For editorial comment, see p 1079. the precedent hypothesis. (Internal Medicine 41: 1124-1129, 2002) Subjects and Methods Key words: CPT, gender difference, diabetic neuropathy, dia- betes This study was a part of a large longitudinal study on aging (National Institute for Longevity Sciences - Longitudinal Study on Aging; NILS-LSA). The design of the study and its objectives have been published previously (7). In brief, par- ticipants of this study were recruited from stratified random From the Department of Public Health, Nagoya City University Medical School, Nagoya and *the Department of Epidemiology, National Institute for Longev- ity Sciences, Obu Received for publication June 1, 2001; Accepted for publication June 13, 2002 Reprint requests should be addressed to Dr. Kiyoshi Takekuma, the Department of Public Health, Nagoya City University Medical School, 1 Kawasumi, Mizuho-cho, Mizuho-ku, Nagoya 467-8601

1124 Internal Medicine Vol. 41, No. 12 (December 2002) Hyperesthesia and Diabetes

samples of community-dwelling people (aged 40 to 79 years tudes until a minimal, but consistent threshold was detected. old) who agreed to participate in Obu-shi and Higashiura-cho, One CPTvalue corresponded to 10 |nA. All measurements of located in a suburb of Nagoya-shi, Japan. Detailed examina- CPTvalues for a single participant were carried out at one time tions including medical evaluation, sensory aging evaluation, within an average period of about 5 minutes. The intra- and body composition analyses, anthropometry, physical function interrater reliability in this method have been described else- evaluation, nutritional analyses, and psychological tests were where (1 1). conducted. The subjects in the present study were participants Statistical analyses were performed with SAS statistical of the examination of NILS-LSAfrom November1997 to package release 6.12 (SAS Institute Inc., Cary, NC, 1997). March 2000. Umezawaand coworkers defined hyperesthesia at a lower value Total number of participants in this study were 2,074 (male: less than mean -1 standard deviation (S.D.), and hypoesthesia 1,050, female: 1,024). Detailed information on the examina- whenthe CPTvalue was higher than mean+2S.D. in non-dia- tion in NILS-LSA can be found at: http: //www.nils.go.jp/in- betic subjects (12). Along with this definition, we calculated dex-e.html. In the present study, the results of the question- the reference CPTvalue for hyper- and hypoesthesia, by using naire, blood pressure, body mass index (BMI), CPT test, and the data from subjects, whose FPGwas lower than 110 mg/dl chemical analyses, such as fasting plasma glucose (FPG), and HbAlc was lower than 6.5%, without a history of diabetes glycated hemoglobin (HbAlc), fasting plasma insulin (FPI), total mellitus. With these reference values, CPTcategories at three cholesterol (TC), triglyceride (TG), and HDL-cholesterol frequencies in each subject were determined. Comparisons of (HDL-C), were used for statistical analyses. In addition, we characteristics amongthe three categories were done with analy- calculated the Homeostasis Model Assessment (HOMA)in- sis of variance by the GLMprocedure in continuous variables, dex as follows: HOMAindex= (FPI) x (FPG)/405. If this in- or with the Cochran-Mantel-Haenzel test (CMHtest) by the dex was 1.73 or more, then insulin resistance was defined to FREQprocedure in categorical variables. After excluding sub- exist (8). jects with hypoesthesia at each frequency, the rates of hyperes- Total subjects were divided into three categories according thesia in the three categories were also calculated, and the trend to glucose metabolic status (normal, insulin resistance, and between prevalence of hyperesthesia and glucose metabolic diabetes) in each gender. Diabetes mellitus was defined when categories was analyzed by CMHtest controlled for age. We their HbAlc was >6.5%, which corresponds to the diagnostic performed age-adjusted multivariate logistic regression analy- criteria of diabetes of The Japan Diabetes Society (9), or they ses for investigating the associations between hyperesthesia and were treated as diabetes mellitus. Thenthe remaining subjects glucose metabolic status in each gender. In these logistic analy- were categorized as normal or insulin resistance according to ses, subjects with hypoesthesia were also excluded. CPTvari- HOMAindex. able was considered as the dependent variable coded 0 (nor- The definitions of other diseases were as follows; hyperten- mal CPT) and 1 (hyperesthesia), and glucose metabolic cat- sion (HT) was defined when they were treated, or systolic blood egories were used as independent variables. pressure was >140 mmHgor diastolic blood pressure was >90 Statistical significance was taken as p<0.05. mmHg.Hypercholesterolemia (HCHOL)was defined when a subject showed a TC level of 220 mg/dl or more. Hypertrigly- Results ceridemia (HTG) was defined when the subjects showed a TG level of 150 mg/dl or more. Hypo-HDLcholesterolemia The reference CPTvalues for hyper- and hypoesthesia at (hypoHDL) was defined when a level of HDL-Cwas lower three frequencies in both genders are shown in Table 1. By than 40 mg/dl. Subjects who drank 30 ml of ethanol (720 ml of using these reference values, CPTvalues of subjects were cat- beer or 300 ml of wine, or 60 ml of 50% whisky) or more per egorized as hypoesthesia, normal perception, or hyperesthesia day were defined as having a drinking habit. at each frequency. Table 2 shows the distribution of CPT val- The CPTtest is used to evaluate three different types of pe- ues and the prevalence of hyper- and hypoesthesia in total sub- ripheral sensory nerve fibers (Ap, A8,and C fiber) quantita- jects. The numbersof males with hypoesthesia at the three fre- tively by stimulating three kinds of current frequency (2,000, quencies (2,000, 250, and 5 Hz) were 45, 42, and 43, respec- 250, and 5 Hz) (6). The CPTtesting was performed in a rela- tively quiet roomwith the subject seated. The roomwas main- tained at a comfortable temperature by air conditioner. The Table 1. Reference CPT Values for Hyper- and Hypoesthesia stimulus site was located on the left index finger. The device M a le s (n = 7 7 4 ) F e m a le s (n = 8 3 8 ) emits sinusoidal alternating currents at 2,000, 250 and 5 Hz at intensities from 0 to 10 mA. This constant current output auto- M e a n S . D . H y p e r H y p o M e a n S . D . H y p e r H y p o matically compensates for alterations in skin resistance and provides a standardized stimulus independent of different skin 2 , 0 0 0 H z 2 3 9 6 2 1 7 7 > > 3 6 3 2 4 0 6 3 1 7 7 > > 3 6 6 thickness, degree of skin dryness or , or drying of 2 5 0 H z 9 9 4 6 5 3 > > 19 1 9 2 4 4 4 8 > > 1 8 0 the electrode paste (10). Electrical stimulus was initially in- 5 H z 6 6 4 1 2 5 > > 14 8 5 4 3 5 1 9 > > 1 2 4 creased until a specific sensation wasreported by the subject. Short stimuli were then applied at progressively lower ampli- *S.D. : standard deviation, Hyper: hyperesthesia, Hypo: hypoesthesia.

Internal Medicine Vol. 41, No. 12 (December 2002) 1125 Takekumaet al tively, and in females, 36, 36, and 34, respectively. in both genders. The subjects with hypoesthesia were excluded Table 3 shows the characteristics of total subjects according in the subsequent analyses. In males, there were statistically to glucose metabolic status in both genders. Male and female significant trends between impaired glucose metabolic status subjects with diabetes were older and had a higher prevalence and the prevalence of hyperesthesia at 2,000 and 250 Hz. In of hypertension than the rest of the groups. Onthe other hand, females, the same trends were observed at 2,000 and 250 Hz. the patterns of associations amongthe three categories regard- Tests for trends in hyperesthesia at 5 Hz regarding glucose ing the level of BMI or prevalence of dyslipidemia differed metabolic impairment showedno statistical significance in ei- between the genders. ther gender. Table 4 shows the prevalence of hyperesthesia at each fre- Since the diabetic groups were of higher age than the other quency in subjects according to glucose metabolic categories groups in both genders, we conducted multivariate logistic re- gression analysis controlled for age (Table 5). In this analysis, hyperesthesia at 2,000 and 250 Hz were significantly associ- ated with insulin resistance in males. In addition, the odds ra- Table 2. Distribution of CPTValues and Prevalence of Hyper- and Hypoesthesia in Each Gender tio for hyperesthesia at 250 Hz in male diabetic subjects was significantly high in this model. In females, significantly high Male (n=1,050) odds ratios for hyperesthesia at 2,000 and 250 Hz were seen in M e a n + S . D . M ax M in. H y pe^ H yp ot diabetic subjects. Insulin resistance was not associated with hyperesthesia at any frequency in females in this model. Hy- 2 , 0 0 0 H z 2 4 0 ｱ 6 2 5 15 5 4 13.2 (139) 4.3 (45) peresthesia at 5 Hz was not associated with diabetes mellitus 2 5 0 H z 9 9 ｱ 4 5 4 36 4.5 12.0 (126) 4.0 (42) or insulin resistance in either gender in this model. 5 H z 6 5 ｱ 3 9 44 4 2 9.4 (99) 4.1 (4 3) In addition, wealso performed logistic regression analyses excluding patients whohad been treated for diabetes (Table 5). These analyses showed higher odds ratios for hyperesthesia at Female (n=1 ,024) 2,000 and 250 Hz in male and female diabetic subjects com-

M e a n ｱ S . D . M ax. M in. H yp ei* H yp o ^ pared with corresponding groups in the previous analyses.

2 , 0 0 0 H z 2 4 0 ｱ 6 3 5 30 2 9 14.0 (143) 3.5 (36) Discussion 2 5 0 H z 9 2 ｱ 4 5 4 13 5.1 14.4 (147) 3.5 (36 ) 5 H z 5 4 ｱ 3 5 3 3 8 3.1 12.0 (123) 3.3 (34) In the present study, we examined the association between hyperesthesia and glucose metabolic impairment in commu- f% (number of subjects). *S.D.: standard deviation, Max.: maximum, nity-dwelling people. Our results indicate that the trends be- Min.: minimum, Hyper: hyperesthesia, Hypo: hypoesthesia. tween glucose metabolic status and the prevalence of hyperes- thesia at 2,000 and 250 Hz were significant in both genders in

Table 3. Characteristics of Subjects Subgrouped into Three Groups According to Glucose Metabolic Status in Both Genders M a le s F e m a le s

N IR D P v a lu e N IR D P v a l ue

N u m b e r 4 7 2 4 8 6 9 2 4 5 2 5 15 5 7 A g e 5 8 . 6 + l l . 1 5 8 . 8 + 1 0 . 9 6 2 . 8 + 9 . 5 < 0 . 0 0 1 5 7. 5+ 1 1 .0 5 9 . 9 + 1 0 . 8 6 5 . 0 + 9 . 1 <0 . 0 1 B M I 2 1 . 7 + 2 . 4 2 4 . 1 + 2 . 6 2 3 . 2 + 3 . 2 < 0 . 0 1 2 1 . 5 +2 . 5 2 3 . 9 + 3 . 4 2 4 . 8 + 3 . 9 <0 . 0 1 F P G ( m g / d l ) 9 6 + 8 . 3 1 0 6 + 1 4 . 6 1 5 7 + 4 4 . 9 <0 .0 0 1 9 2 + 7 . 3 1 0 2 + 1 2 . 9 1 5 5 + 4 5 . 7 < 0 . 0 0 1 H b A lc (% ) 5 . 1 + 0 . 4 5 . 2 + 0 . 4 7 . 2 + 1 . 3 < 0 . 0 0 1 5. 0 + 0. 3 5 . 2 + 0 . 4 7 . 1 + 1 . 4 < 0 . 0 0 1 S m o k in g (% ) 4 2 . 4 3 4 . 4 3 8 . 0 n . s . 8. 6 5 . 8 3 . 5 < 0. 0 5 D r i n k i n g ( % ) 3 9 . 4 3 6 . 0 3 1 . 5 n .s. 3. 3 4 . 3 3 . 5 n .s. H T ( % ) 2 1. 0 3 4 .6 4 1 .3 < 0 . 0 1 1 9. 3 3 8 . 5 6 6 . 7 <0 . 0 1 H C H O L ( % ) 3 7 . 5 4 4 . 2 4 1 . 3 n. s. 5 1. 1 6 1 . 8 5 4 . 4 < 0. 0 5 H T G ( % ) 1 6 . 5 3 8 . 3 3 5 . 9 < 0 . 0 1 8. 6 2 4 . 9 3 3 . 3 <0 . 0 1 h y p o H D L ( % ) 5 . 1 l l . 5 5 . 4 < 0 . 0 5 0. 9 1 . 6 3 . 5 n . s .

P value from analysis of variance was used for continuous variables; Cochran-Mantel-Haenzel test (CMHtest) was used for categorical variables. These analyses were used to test for overall differences amongthree groups. Data are means±standard deviation or percentage, n.s.: not significant, N: normal, IR: insulin resistance, D: diabetes HCHOL:hypercholesterolemia, HTG: hypertriglyceridemia, hypoHDL: hypo HDLcholesterolemia.

1126 Internal Medicine Vol. 41, No. 12 (December 2002) Hyperesthesia and Diabetes

Table 4. Prevalence of Hyperesthesia According to Diabetic Category in Both Genders Males

N IR D P f or t re nd *

2 , 0 0 0 H z 1 1. 0 (49 /4 47) 1 6 .5 (7 8 /4 7 3) 1 4 . 1 (1 2 /8 5 ) P < 0 . 0 5 2 5 0 H z 9. 4 ( 4 2/ 44 6 ) 1 4 .8 (7 0 /4 72 ) 1 5. 6 ( 14 /9 0) P < 0 .0 5 5 H z 9 .3 (4 2 /4 5 1) 1 0 .8 (5 0 /4 65 ) 7. 7 (7 /9 1) n .s .

Fe ma l e s

N IR D P fo r t re nd *

2 , 0 0 0 H z 1 3. 1 (5 7 /4 3 6 ) 1 4. 9 ( 7 4/ 4 98 ) 2 2 .2 (1 2/ 5 4) / > <0 . 05 2 5 0 H z 1 4. 0 ( 6 1/ 4 35 ) 1 4. 6 ( 7 3/ 5 00 ) 2 4 .5 (1 3/ 5 3) P < 0 . 0 5 5 H z l l.7 (5 1/4 3 6 ) 1 2. 6 ( 6 3/ 4 99 ) 1 6. 4 (9 /5 5) n . s .

%(number: subjects with hyperesthesia/total subjects excluding subjects with hypoesthesia). N: normal, IR: insulin resistance, D: diabetes, n.s.: not significant. *P for trend was assessed with CMHtest controlled for age.

CMHtest controlled for age. In addition, there were signifi- Table 5. Odds Ratio (OR) and 95% Confidence Interval (CI) of cantly high odds ratios for hyperesthesia at 2,000 and 250 Hz Hyperesthesia at Three Frequencies for Those Whohad Insulin Resistance or Diabetes in female diabetic subjects, and at 250 Hz in male diabetic M ales F em ale s subjects in age-adjusted logistic regression analyses. These O R * ( 9 5 % C I ) O R * ( 9 5 % C I ) results could be interpreted to indicate that hyperesthesia de- velops in association with occurrence of glucose metabolic A ll su bjects impairment in a community-dwellingpopulation. 2 , 0 0 0 H z N r e f e r e n c e r e f e r e n c e Dyckand coworkers reported that hyperesthesia is notewor- IR 1 .64 ( 1. ll -2 .4 1)* 1. 25 ( 0. 86 -1 .83 ) thy as an initial symptomof diabetic polyneuropathy (3). Ani- D 1 .5 6 (0 . 78 -3 . ll ) 2 .4 2 (1 .1 8 -4 .9 7 )* mal studies have demonstrated that Wallerian degeneration with 2 5 0 H z N r e f e r e n c e r e f e r e n c e macrophage influx is associated with the development of hy- IR 1 .71 ( 1. 14 -2 .5 8)* 1. 14 ( 0. 79 -1 .65 ) peresthesia (13). Although there has been no pathological in- D 2 .0 9 (1 . 07 -4 . 05 )* 2 .6 5 (1 . 31 -5 . 37 )* * vestigation into the hyperesthetic state in humanbeings, the 5 H z N r e f e r e n c e r e f e r e n c e present results and the available evidence support the notion IR 1 .17 ( 0. 76 -1 .8 1) 1. 09 ( 0. 74 -1 .62 ) D 0 .7 9 (0 .34 - 1.83) 1.4 8 (0 .6 8- 3 .24 ) that hyperesthesia in the CPTtest maybe an initial sign of peripheral nerve impairment in diabetic patients. Subjects excluded patients who had been treated as diabetes*** In one clinic-based study using the CPTtest, Umezawaand 2 , 0 0 0 H z N r e f e r e n c e r e f e r e n c e coworkers reported that approximately 40%of diabetic patients IR 1 .6 4 (1. ll -2 .4 1) * 1. 25 ( 0. 86 -1 .8 3) without overt neuropathy suffer from hyperesthesia ( 1 2). Com- D 3. 66 ( 1.5 1- 8. 88 )* * 2. 81 ( 0. 85 -9 .2 4) pared with their result, the prevalence of hyperesthesia in dia- 2 5 0 H z N r e f e r e n c e r e f e r e n c e betic subjects waslower in the present study. In our study, the IR 1.7 1 ( 1. 14- 2.5 8)* * 1.1 4 (0. 79- 1.6 5) rates of hyperesthesia in diabetic subjects ranged from 14 to D 2 .9 2 (1 . ll -7 .7 0 )* 4. 90 (1 .6 5- 14 . 56 )* * 22%at 2,000 Hz, and from 16 to 25%at 250 Hz. This discrep- 5 H z N r e f e r e n c e r e f e r e n c e ancy could be explained by the difference in research design IR 1 .1 7 (0 .7 6-1 .8 1) 1 .0 9 ( 0. 74 -1 .6 2) and CPTmeasurement between these two studies. One pos- D 1. 15 (0 .3 3- 3.9 8 ) 1.7 5 (0.4 8- 6 .38 ) sible reason is the difference in subjects' characteristics be- tween these two studies. In the present study, participants were N: normal, IR: insulin resistance, D: diabetes, ^controlled for age. recruited from a community-based population. It is possible *: P<0.05, **: P<0.01, ***: 63 male and 41 female diabetic patients that they might suffer diabetic complications to a lesser extent were excluded. than those from a clinic-based population. Another possible reason is that the location site of electrodes might have affected the prevalence of hyperesthesia in the dia- betic subjects in this study. Umezawaand coworkers measured

Internal Medicine Vol. 41, No. 12 (December 2002) 1127 Takekumaet al the CPTvalues at the toes (12), whereas we measured CPT sociation between insulin resistance and hyperesthesia at those values on the index fingers. In case of minimal symptoms or frequencies in females. These results were compatible with the signs of diabetic neuropathies, the distal lower extremity re- notion that males suffer from diabetic neuropathy earlier in sponses are more commonlyabnormalthan upper extremity the course of diabetes than females. responses by electrodiagnostic examination ( 1). Thus, measure- In multivariate logistic analyses, there wasno significant ment at the finger maybe related to the lower prevalence of odds ratio at 2,000 Hz in male diabetic subjects, on the other hyperesthesia in diabetes in this study. hand, a significant odds ratio wasseen in female diabetic sub- In the present study, there was no relationship between im- jects. These results could be also explained by gender differ- paired glucose metabolic status and hyperesthesia at 5 Hz, ence in the course of development of diabetic neuropathy. Pre- which is associated with C fiber function, in both genders. C vious studies reported that patients with progressive state of fibers have a smaller nerve diameter than A(3 and A8fibers. In diabetic neuropathy showed significantly higher CPT values, diabetes mellitus, smaller fibers are moreprone to be affected which indicate hypoestheia, compared with normal subjects than larger fibers (14, 15). Umezawa and coworkers reported (21). It is conceivable that diabetic patients may show pseudo that hyperesthesia at 5 Hz wasassociated with diabetes melli- normal CPTvalues for a time in the course from hyperesthesia tus in their clinic-based study (12). to hypoesthesia with the deterioration of diabetic neuropathy. There are two possible explanations for this inconsistency In this study, wealso calculated age-adjusted logistic regres- between our results and the previous knowledge. The first is sion analyses excluding subjects whohad been treated as dia- due to a characteristic of CPTtest at 5 Hz, which shows larger betes to eliminate subjects with a long history of the disease. variability in CPT values than other frequencies. As previously In these analyses, the odds ratio at 2,000 Hzin male diabetic described, the coefficient of variation of CPTmeasures was subjects was 3.66 (95% CI, 1.51 to 8.88) with statistical sig- frequency dependent, with the smallest coefficient of variation nificance. The corresponding odds ratio in females was 2.81 at 2,000 Hz and the largest at 5 Hz (16). To clarify the associa- (95% CI, 0.85 to 9.24), which was higher than the previous tion between hyperesthesia at 5 Hz and glucose metabolic im- method but not statistically significant due to the reduced num- pairment in a community-based population, a larger number ber of subjects. It would be logical to speculate that the non- of subjects is necessary to obtain adequate statistical power significant odds ratio at 2,000 Hz in male diabetic subjects in compared with other frequencies. The lack of a sufficient num- the previous method might be attributed to the increase of CPT ber of diabetic subjects in our study might lead to non-signifi- values in diabetic patients with a long duration of the disease. cant association between hyperesthesia at 5 Hzand impaired Thus, the observed gender difference of the odds ratios at 2,000 glucose metabolic status. Hz in diabetic groups may be ascribed to the gender dissimi- The second possibility is that the location site of electrodes larity in the course of development of diabetic neuropathy. might affect the results. There is a significant difference in CPT There are somelimitations in this study. First, the notion, values between the upper and lower limbs (17). The difference that hyperesthesia detected by the CPT test may be the initial of location site might also influence this inconsistency. symptom of nerve fiber impairment, is not based on pathologi- In the present study, there seems to be a gender difference cal findings. There has been no pathological report regarding in the association between hyperesthesia and impaired glucose hyperesthesia by CPTtest in human beings. In addition, abil- metabolic status. In logistic regression analyses, significantly ity of CPTtest to measure selective sensory nerve fiber func- high odds ratios of hyperesthesia at 2,000 and 250 Hz in fe- tion abnormality has also not been validated by pathological male subjects with diabetes were seen in the age-adjusted examination. Thus, the results of this study should be cautiously model. On the other hand, in males, insulin resistance was as- interpreted until pathological examination is used to clarify sociated with hyperesthesia at those frequencies, but diabetes the meaning of hyperesthesia detected by CPT test in diabetic was only associated with hyperesthesia at 250 Hz. patients. The second limitation is the lack of precise neuro- While little is known about the development of hyperesthe- logical examination for the participants. To distinguish hyper- sia in the course of diabetes mellitus, several epidemiologic esthesia in the normal group from that in the diabetic group, studies reported that males are moreprone to suffer diabetic another quantitative sensory examination is needed. Further neuropathy compared with females (4, 18, 19). Additionally, investigation should be carried out to clarify this issue. The one report indicated the possibility that males experience sen- third limitation is the validity of CPTvalues in determining sory disturbance earlier in the course of diabetes compared with the criteria of hyperesthesia in both genders. In this study, we females (18). Taking into account the hypothesis that hyperes- defined hyperesthesia at a lower value, less than mean-1 S.D., thesia may be an indicator of diabetic neuropathy, it seems in non-diabetic subjects. However,the CPTvalues in the three reasonable that the gender difference in susceptibility for dia- frequencies, especially at 250 and 5 Hz, in both genders did betic neuropathy mayaffect the results of the present study. not show normal distribution, and they might lead to the ob- Since insulin resistance is knownto be a risk factor for dia- served gender differences in the prevalence of hyperesthesia in betes (20), it is possible that there exist mild diabetic subjects normal, insulin resistance, or diabetic groups. To determine the in insulin resistance groups. In males, multivariate logistic appropriate CPTvalues for hyperesthesia, another method might analysis showedthat insulin resistance wasassociated with be employed. Thefourth limitation is that the research design hyperesthesia at 2,000 and 250 Hz, however, there was no as- in this study wascross-sectional. In the cross-sectional study

1128 Internal Medicine Vol. 41, No. 12 (December 2002) Hyperesthesia and Diabetes

the National Institute for Longevity Sciences, Longitudinal Study of Ag- model, we could not determine whether glucose metabolic ing (NILS-LSA). J Epidemiol 10: S 1-S9, 2000. impairment emerged prior to hyperesthesia by CPT test. How- 8) Oimatsu H, Saitoh S, Ura N, Shimamoto K. A practical index for evalua- ever, the tests for trends in hyperesthesia at 2,000 and 250 Hz tion of insulin resistance. Journal of the Japan Diabetes Society 43: 205- regarding glucose metabolic impairmentwere significant, sup- 213, 2000 (in Japanese, Abstract in English). porting the notion that hyperesthesia develops accompanied by 9) KuzuyaT, NakagawaS, Satoh J, et al. Report of the committee of Japan glucose metabolic impairment. A longitudinal study is needed Diabetes Society for the diagnostic criteria for diabetes mellitus. Journal to address this relationship. of the Japan Diabetes Society 42: 385-404, 1999 (in Japanese, Abstract in English). The present study provides new information about the rela- 10) Katims JJ, Patil AS, Rendell M, et al. Current perception threshold screen- tionship between hyperesthesia with CPTtest and glucose ing for . Arch Environ Health 46: 207-212, 1991. metabolic status. First, hyperesthesia at 2,000 and 250 Hz were 1 1) Takekuma K, Ando F, Niino N, Shimokata H. Age and gender differences associated with impaired glucose metabolic status in a com- in skin sensory threshold assessed by current perception in community- dwelling Japanese. J Epidemiol 10: S33-38, 2000. munity-based population. Second, our result suggested that a 12) Umezawa S, Kanamori A, Yajima Y, Aoki C. Current perception thresh- gender difference might exist in the association between hy- old in evaluating diabetic neuropathy. Journal of the Japan Diabetes So- peresthesia and glucose metabolic impairment. Recently, one ciety 40: 711-718, 1997 (in Japanese, Abstract in English). study reported that hyperesthesia occurs in mild diabetic poly- 13) Myers RR, YamamotoT, Yaksh TL, Powell HC. The role offocal nerve neuropathy, suggesting it might be attributed to impairment of and Wallerian degeneration in peripheral nerve injury produc- ing hyperesthesia. Anesthesiology 78: 308-316, 1993. diabetic polyneuropathy (3). Further study is necessary to ex- 14) Mattingly GE. Fischer VW. following prolonged amine the possibility that hyperesthesia as revealed by the CPT exposure to streptozotocin-induced diabetes in rats: a teased nerve fiber test maybe a useful marker of diabetic polyneuropathy. study. Acta Neuropathol 59: 133-138, 1983. 15) Jensen TS, Bach FW, Kastrup J, Dejgaard A, Brennum J. Vibratory and References thermal thresholds in diabetics with and without clinical neuropathy. Acta Neurol Scand 84: 326-333, 1991. 1) Consensus report. Quantitative sensory testing: A consensus report from 16) Masson EA, Veves A, Fernando D, Boulton AJ. Current perception thresh- the peripheral neuropathy association. 43: 1050-1052, 1993. olds: a new, quick, and reproducible methodfor the assessment of pe- 2) Verdugo R, Ochoa JL. Quantitative somatosensory thermotest. A key ripheral neuropathy in diabetes mellitus. Diabetologia^2: 724-728, 1989. method for functional evaluation of small calibre afferent channels. Brain 17) Katims JJ, Naviasky EH, Rendell MS, Ng LK, Bleecker ML. Constant 115: 893-913, 1992. current sine wavetranscutaneous nerve stimulation for the evaluation of 3) Dyck PJ, Dyck PJ, Larson TS, O'Brien PC, Velosa JA. Patterns ofquanti- peripheral neuropathy. Arch Phys Med Rehabil 68: 210-213, 1987. 1 8) Harris M, Eastman R, Cowie C. Symptomsof sensory neuropathy in adults tative sensation testing of hypoesthesia and are predictive of with NIDDMin the U.S. population. Diabetes Care 16: 1446-1452, 1993. diabetic polyneuropathy: a study of three cohorts. Nerve growth factor study group. Diabetes Care 23: 510-517, 2000. 19) Franklin GM, Kahn LB, Baxter J, Marshall JA, Hamman RF. Sensory neuropathy in non-insulin-dependent diabetes mellitus. The San Luis Val- 4) The DCCTResearch Group. Factors in development of diabetic neuropa- ley Diabetes Study. AmJ Epidemiol 131: 633-643, 1990. thy. Baseline analysis of neuropathy in feasibility phase of Diabetes Con- trol and Complications Trial (DCCT). Diabetes 37: 476-481, 1988. 20) Ohmura T, Ueda K, Kiyohara Y, et al. The association of the insulin resis- 5) Said G, Goulon-Goeau C, Slama G, Tchobroutsky G. Severe early-onset tance syndrome with impaired glucose tolerance and NIDDMin the Japa- nese general population: the Hisayama study. Diabetologia. 37: 897-904, polyneuropathy in insulin-dependent diabetes mellitus. A clinical and 1994. pathological study. N Engl J Med 326: 1257-1263, 1992. 6) Liu S, Kopacz DJ, Carpenter RL. Quantitative assessment of differential 21) Rendell MS, Dovgan DJ, Bergman TF, O'Donnell GP, Drobny EP, Katims JJ. Mapping diabetic sensory neuropathy by current perception threshold sensory60-63, nerve1995.block after spinal anesthesia. Anesthesiology 82: testing. Diabetes Care 12: 636-640, 1989. 7) Shimokata H, Ando F, Niino N. A new comprehensive study on aging -

Internal Medicine Vol. 41, No. 12 (December 2002) 1129