Effect of Aldose Reductase Inhibition on Heart Rate Variability in Patients with Severe or Moderate...

Effect of Aldose Reductase Inhibition on Heart Rate Variability in Patients with Severe or Moderate DiabeticAutonomic NeuropathyTriantaffilos P. Didangelos,1 Vasilios G. Athyros,2 Dimitrios T. Karamitsos,1 Athanasios A. Papageorgiou,3 Georgios I. Kourtoglou1 and Athanasios G. Kontopoulos2

1 Division of Diabetology, Aristotelian University, Hippocration Hospital, Thessaloniki, Greece2 Division of Cardiology, Aristotelian University, Hippocration Hospital, Thessaloniki, Greece3 2nd Propedeutic Department of Internal Medicine, Aristotelian University, Hippocration Hospital,

Thessaloniki, Greece

Summary Patients with diabetic autonomic neuropathy (DAN) have an increased cardio-vascular mortality rate compared with diabetic patients without DAN. Heart ratevariability (HRV) time and frequency domain indices are strong predictors ofmalignant arrhythmias and sudden cardiac death. This prospective, randomised,double-blind, placebo-controlled study analysed the long-term effect of an aldosereductase inhibitor, tolrestat, on HRV time and frequency domain variables in 45patients with diabetes mellitus (DM) and DAN. Patients were randomised intotolrestat (n = 22) and placebo (n = 23) groups. Tolrestat (200 mg/day) or placebowere administered, respectively, for a period of 12 months. HRV was assessed atmonths 0, 3, 6, 9 and 12. The HRV level of the 45 patients was compared withthat of 20 patients with DM, with analogous glycaemic control, without DANand 20 healthy controls, of similar age and gender. At the twelfth month, tolrestat,compared with placebo, had a beneficial effect on HRV indices related to vagaltone. Compared with baseline, HRV time and frequency domain indices showedno significant improvement. Moreover, at the twelfth month of tolrestat admin-istration, HRV indices remained less than that of patients with DM but withoutDAN, and healthy controls. The 12 patients of the 22 with moderate DAN bene-fited more than the 10 patients of the 22 with severe DAN. At the twelfth monthno patient showed deterioration in HRV indices with tolrestat as was seen withplacebo. Our data suggest that tolrestat slows down the progression of DANcompared with placebo. This effect of an aldose reductase inhibitor may contrib-ute to a reduction in risk for malignant ventricular arrhythmias. The early detec-tion of DAN is imperative for successful intervention.

CLINICAL PHARMACODYNAMICS Clin. Drug Invest. 1998 Feb; 15 (2): 111-1211173-2563/98/0002-0111/$05.50/0

© Adis International Limited. All rights reserved.

Experimental evidence for an association be-tween a propensity for lethal arrhythmias and signsof reduced vagal and/or increased sympatheticactivity has encouraged the development of quan-titative markers of autonomic activity. Heart ratevariability (HRV) is one of the most reliable of suchmarkers.[1] Decreased HRV level is a strong predic-tor of arrhythmic events, sudden death and totalcardiovascular mortality in patients with auto-nomic system impairment.[1-3] HRV level is a reli-able tool for assessing diabetic autonomic neuro-pathy (DAN).[4]

Patients with DAN exhibit a high cardiacmortality rate in comparison with patients withdiabetes mellitus (DM) without DAN.[5-7] Thisfact, combined with the increased prevalence ofDAN in patients with diabetes for a period greaterthan 10 years,[6] indicates the significance ofDAN prevention, mainly with effective glycaemiccontrol. However, total prevention of neuropathyis not achievable.[8] This was shown in the DiabetesControl and Complications Trial[9] for patientswith insulin-dependent DM (IDDM) and in a studyby Ohkubo et al.[10] for lean Japanese patients withnon−insulin-dependent DM (NIDDM). Thus, thetimely detection of DAN and the use of effectivemeans to improve autonomic function or to slowdown the progression of DAN become of para-mount importance.[8,11]

The present prospective, randomised, double-blind, placebo-controlled study was undertakento investigate the long-term effect of tolrestat, analdose reductase inhibitor (ARI), in comparisonwith placebo on HRV level in patients with DAN.

Patients and Methods

Study Participants

Forty-five patients (20 men, 25 women) with anage range from 19 to 70 years (median 52 years)were studied. The inclusion criteria were DM andDAN. The latter was established if two or morestandard cardiovascular reflex tests of autonomicfunction were abnormal. Expiration-inspirationratio, standard deviation and mean circular resul-

tant of R-R intervals of the deep breathing test,Valsalva manoeuvre, postural index and posturalhypotension[12,13] were used. If two of the cardio-vascular reflex tests were abnormal the patient wasconsidered to have moderate DAN, and if three ormore of these tests were abnormal the patient wasconsidered to have severe DAN.

The exclusion criteria were coronary arterydisease, heart failure, cardiomyopathies, arterialhypertension, atrial fibrillation, bundle branchblock, treatment with an ARI, a β-blocker or anACE inhibitor at the time of the run-in phase, andunwillingness to participate in the study.

Patients were considered free of coronary arterydisease on clinical and non-invasive criteria. 20patients with DM without DAN, with analogousglycaemic control, and 20 healthy volunteers wereused as control groups. The control group, of simi-lar age and gender to the patients, were examinedat the same time as the patients.

Informed consent, which was approved by theEthics Committee of the Hippocration Hospital,was obtained from all patients before enrolment.Basic characteristics of the study population arepresented in table I.

Study Protocol

Patients with DAN and controls underwentbaseline ambulatory 2-channel 24-hour electro-cardiogram monitoring. HRV time and frequencydomain indices were assessed. Patients withDAN were then randomised into either thetolrestat (n = 25) or the placebo group (n = 25).Tolrestat (200 mg/day) or placebo were adminis-tered to patients in the two groups, respectively.Consecutive 24-hour electrocardiogram monitor-ings to assess HRV were carried out in patients andcontrols 3, 6, 9 and 12 months later, while patientswere still receiving the same medical management.HRV analyses were performed blindly by one ofthe investigators (VGA).

Echocardiography

To confirm normal structure and function of theleft ventricle and to investigate any possible effects

112 Didangelos et al.

© Adis International Limited. All rights reserved. Clin. Drug Invest. 1998 Feb; 15 (2)

of tolrestat on them, echocardiograms were ob-tained before each HRV assessment as describedpreviously.[14] The Sigma 1AC (Kontron Instru-ments, Montigny, France) echocardiograph wasused.

HRV Analyses

The Cardiolight System (Medset, Hamburg,Germany) was used. This consists of a solid statedigital electrocardiogram recorder and an IBM(Thessaloniki, Greece) 486 DX2/66 MHz com-puter, working on the Medset software. This pro-gram automatically detects abnormal QRS com-plexes and filters them out, so that the HRV iscalculated only on ‘normal’ R-R intervals. Theinterval tachogram is calculated from a sample of1024 R-R intervals. Time accuracy of the recordingand the ability to detect abnormal QRS beats havebeen validated previously.[15,16]

The programme produced the following timedomain indices:• Standard deviation of all 24-hour R-R intervals

(SDNN/24h) in msec• Mean of standard deviations of R-R intervals of

all 5-minute segments (SDNN/5 min) in msec• Standard deviation of the average R-R intervals

of all 5-minute segments (SDANN/5 min) inmsec

• Percentage of the R-R intervals differing morethan 50msec of all 2-minute segments (pNN50)

• Root-mean-square of the differences in succes-sive R-R intervals of all 2-minute segments(RMSSD) in msec. The first three indices express total HRV (long

and short time), while the last two reflect vagaltone.

In addition, the following frequency domainanalysis indices were obtained using the fast-Fourier transform:• Total power (0-0.4Hz)• High frequency power (0.15 to 0.4Hz) spectral

density• Low frequency power (0.04 to 0.15Hz) spectral

density and• Very low frequency power (0.0 to 0.04Hz) spec-

tral density. Total power expresses the magnitude of the en-

tire HRV, high frequency power reflects the para-sympathetic tone, and low frequency power thesympatho-vagal interactions. Total power and frac-tional power spectral density are expressed in ab-solute units (msec2).

Moreover, high frequency power and low fre-quency power are expressed in normalised units;this is accomplished by dividing the power (inmsec2) of the given component by the total powerminus the very low frequency component and then

Table I. Baseline characteristics of the study population.[14] Values are expressed as means ± standard error, unless otherwise defined

Group Tolrestat (n = 22)

Placebo (n = 23)

DM without DAN (n = 20)

Healthy controls (n = 20)

Age [y] (range) 52 (19-70) 50 (20-68) 51 (20-65) 52 (21-68)

Gender: male/female 9/13 10/13 10/10 9/11

IDDM/NIDDM 11/11 11/12 9/11

Insulin treatment yes/no 22/0 23/0 12/8

Symptomatic/asymptomatic DAN 8/14 8/15

Retinopathy: yes/no 22/0 23/0 2/18 0/20

Overt nephropathy: yes/no 0/22 0/23 0/20 0/20

Known duration of diabetes [y] (range) 17 (9-33) 16 (8-44) 6 (1-18)

HbA1c % (range) 8.5 (7.4-11.3) 8.4 (7.6-10.9) 8.1 (7.8-10.3) 5.5 (4.6-6.1)

Systolic/diastolic blood pressure (mm Hg) 123 ± 2/77 ± 1 122 ± 2/79 ± 1 118 ± 2/76 ± 1 115 ± 2/75 ± 1

Left ventricular ejection fraction (%) 61 ± 1 59 ± 1.1 61 ± 0.9 63 ± 0.9

Abbreviations: DAN = diabetic autonomic neuropathy; DM = diabetes mellitus; HbA1c = glycated haemoglobin; IDDM = insulin-dependentdiabetes mellitus; NIDDM = non−insulin-dependent diabetes mellitus.

Tolrestat and Heart Rate Variability in DAN 113


multiplying by 100. Normalised power combinedwith the low to high frequency ratio provides anassessment of fractional distribution of poweracross the frequency axis, irrespective of the mag-nitude or the modifications of total power,[17] andis highly reliable.

Frequency domain indices were computed hour-by-hour, providing an overall average for the totaltest duration. Spectral plots for each hour were de-rived from 4-minute segments.

Patients with severe neuropathy, establishedby significant decreases in total HRV level plusextremely low time domain parasympathetic vari-ables [which presented a paradoxical prevalence ofhigh frequency power (HFP) over low frequencypower (LFP) despite the notable decreased fre-quency domain parameter values in absolute units]were excluded because their LFP/HFP ratio wasnot reliable.[18]

Follow-Up

Patients were followed on an outpatient basisfor a period of 12 months, with visits at 1-monthintervals. HRV was assessed, as mentioned above,at months 0, 3, 6, 9 and 12. Glycated haemoglobin(HbA1c) was measured every 3 months by the Ion-Capture method, using the IMX Abbott apparatusand the GHb reagent pack (Abbott Laboratories,IL, USA). Normal values were 4.4 to 6.4%. Fullblood count, blood urea nitrogen, serum creatinine,creatine kinase, aspartate aminotransferase, ala-nine aminotransferase, alkaline phosphatase, gammaglutamyl transpeptidase and urinalysis were per-formed every month by standard methods.

Statistical Analysis

Because of the wide and abnormal distributionof values of HRV indices as a result of differencesin severity of DAN, age and gender among thepatients, these are expressed as medians andquartiles (Q1-25%, Q3-75%). Two-way analysis ofvariance, using the Friedman’s rank test (χ2 ap-proximation), was used to compare values. pValues < 0.05 were considered statistically signif-icant. For all forms of statistical analysis the Stat

Graphics 4.0 (Statistical Graphics Corp., Prince-ton, New Jersey, USA) programme was used.

Results

All patients were receiving conventional insulintreatment consisting of two injections of a 30/70mixture of soluble and isophane insulin. Initially,25 patients were randomised into each of the twotreatment groups. Two patients from the placebogroup were lost to follow-up. Thus, 23 remaineduntil the end of the study. Ten of 23 patients fromthe placebo group were considered, according topre-set criteria, to have severe DAN and 13 to havemoderate DAN.

In the tolrestat group, one patient with severeDAN died suddenly 2 months after treatment initi-ation, and 2 patients were withdrawn from thestudy because they presented high transaminaselevels (more than 3-fold above the upper normallimit). Twenty-two patients remained in this groupuntil the end of the study. Ten of 22 patients treatedwith tolrestat were considered to have severe DANand 12 to have moderate DAN.

Standard Cardiovascular Reflex Tests

Baseline and 12-month values of these tests inpatients with DAN receiving tolrestat or placeboare presented in table II. There were no statisticallysignificant differences between the values of thetwo groups at baseline. Expiration/inspiration (E/I)ratio mean value deteriorated significantly in theplacebo group at the twelfth month compared withbaseline (1.04 ± 0.008 vs 1.07 ± 0.008, p < 0.05).The values of standard deviation of beat to beatvariation, E/I ratio and postural hypotension wereimproved with tolrestat. The latter two values weresignificantly better than the corresponding valueswith placebo at the twelfth month, but still less thanthat of controls.

HRV Indices

At baseline, the 24-hour recording analysisshowed that HRV time and frequency domain indi-ces of the patients in the tolrestat and placebo



groups were significantly decreased comparedwith those of patients with DM without DAN andnormal volunteers. High frequency power was pro-portionally more decreased than low frequencypower in patients with DAN. In addition, an ab-sence of circadian variation of HRV with no signif-icant differences between night (2300 to 0700hours) and day (0700 to 2300 hours) was detected.

HRV indices at baseline in the placebo groupwere not statistically different compared withthose of the tolrestat group. The HRV level of con-trols with DM without DAN was slightly, but notsignificantly, lower than that of normal volunteers.Furthermore, the HRV level of both control groupsremained significantly higher than those of theplacebo or the tolrestat group throughout the study.

Patients on placebo demonstrated a deteriora-tion in some HRV indices (RMSSD, HFP innormalised units and LFP/HFP ratio) at the twelfthmonth compared with baseline (p < 0.05). The 22tolrestat-treated patients, as a group, had at thetwelfth treatment month a better HRV level com-

pared with patients on placebo (figs 1 and 2), al-though changes in HRV indices were not signifi-cant compared with baseline. The 12 patients of the22 in the tolrestat group with moderate DAN hadHRV indices that were significantly better thanthose of the 13 patients of the 23 with moderateDAN on placebo, at the twelfth treatment month.However, at that time the indices remained signif-icantly lower than those of both control groups.

At the twelfth treatment month, the 10 patientsof the 22 on tolrestat with severe DAN demon-strated a smaller, marginally significant increase insome time domain HRV indices (pNN50 and RMSSD)compared with 10 of 23 patients on placebo withanalogous severity of DAN. Nevertheless, all HRVindices in these patients remained low in terms ofabsolute values. At the twelfth treatment month, nopatient in the tolrestat-treated group had worseHRV indices values compared with baseline, asseen in the majority of patients on placebo.

The differences among the HRV indices of thetolrestat- and the placebo-treated patients of analo-

Table II. Baseline values (mean ± SD) of cardiovascular tests of autonomic function in the tolrestat and placebo group.[14] Values are expressedas means ± standard error, unless otherwise defined

Group Tolrestat (n = 22) Placebo (n = 23) p value

Expiration/inspiration ratio

Baseline 1.06 ± 0.0089 1.07 ± 0.0083 NS

After 12 months 1.09 ± 0.012* 1.04 ± 0.0081* < 0.05

Standard deviation of beat-to-beat variation

Baseline 19.3 ± 3.1 26.0 ± 4.8 NS

After 12 months 25.8 ± 3.6* 29.8 ± 4.9 NS

Mean circular resultant

Baseline 11.1 ± 2.2 12.9 ± 3.3 NS

After 12 months 14.0 ± 3.1 12.0 ± 3.2 NS

Postural hypotension (mm Hg)

Baseline 20.9 ± 1.1 28.0 ± 0.7 NS

After 12 months 11.8 ± 0.8* 30.1 ± 0.9 < 0.05

Valsalva ratio

Baseline 1.35 ± 0.06 1.32 ± 0.08 NS

After 12 months 1.40 ± 0.08 1.33 ± 0.08 NS

Postural index

Baseline 1.07 ± 0.02 1.08 ± 0.02 NS

After 12 months 1.08 ± 0.02 1.09 ± 0.02 NS

* p < 0.05 vs baseline.Abbreviation: NS = not significant.



gous severity are shown in figures 3 (time domain)and 4 (frequency domain). A trend of circadianHRV pattern towards normal was also observed inmost of the tolrestat-treated patients. Thus, thenight-to-day ratio value of HFP in normalised unitswas increased from 1.1 at baseline to 1.4 at thetwelfth month of treatment (p < 0.05).

The heart rate of patients in the tolrestat groupwas decreased from 80.3 ± 2.1 to 78.4 ± 1.5beats/min (p = NS), while in the placebo group it

increased from 79.6 ± 1.9 to 82.2 ± 2.4 beats/min(p = NS). At the twelfth month, the heart rate in thetolrestat group was not significantly lower than thatof the placebo group. At baseline, no complexventricular arrhythmias were detected in all 45patients with DAN.

Singular premature ventricular beats were morefrequent in the 24-hour recording of these patientscompared with controls. During treatment, patientsin the tolrestat group demonstrated a significant

20

60

100

140

180

220

260

300

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

SDNN/24h

Months

mse

c

n=22 n=23 n=20 n=20

Tolrestat Placebo DM without DAN

NV

0

20

40

60

80

100

120

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

∗

mse

c

Min-Max 25-75% Median value

SDNN/5 min

n=22 n=23 n=20 n=20

∗

−5

5

15

25

35

45

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

pNN50

n=22 n=23

n=20 n=20

∗%

0

20

40

60

80

100

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

Months


NV

RMSSD

n=22 n=23 n=20 n=20

mse

c ∗∗

Fig. 1. Effect of tolrestat on time domain indices of heart rate variability in patients with diabetes mellitus (DM) and diabetic autonomicneuropathy (DAN) compared with placebo and controls. Values are expressed as medians with quartiles (Q1-25%, Q3-75%), plusminimum-maximum value ∗ p < 0.05, and ∗∗ p < 0.01 versus the 12th month of placebo. All time domain indices of both controlgroups were significantly higher than those of the two treatment groups. Abbreviations: NV = normal volunteers; pNN50 = percentageof the R-R intervals differing more than 50msec; RMSSD = root-mean-square of the differences of successive R-R intervals;SDNN/24h = standard deviation of all 24-hour R-R intervals; SDNN/5 min = mean of standard deviations of R-R intervals of all5-minute segments.



reduction of premature ventricular beats (0.8/hourat the twelfth month versus 3.4/hour at baseline,p < 0.001, and versus 3.1/hour in the placebo groupat the twelfth month, p < 0.001). Nevertheless, theyremained higher than those of the control groups(DM without DAN 0.2/hour and normal volunteers0.1/hour, p < 0.01 for both comparisons).

Adverse Effects

Two of the patients presented high values (morethan 3-fold greater than the upper normal limit) ofAST and ALT, at the first and second month of

treatment. Both patients belonged to the tolrestatgroup. They discontinued treatment with tolrestatand were excluded from further participation in thestudy. Transaminase values returned to normalwithin 15 days. No significant change in the valuesof haematocrit, haemoglobin, white blood count orplatelets was recorded.

Glycaemic Control and Arterial Blood Pressure

No significant effect of tolrestat on glycaemiccontrol (HbA1c between 8.3 and 8.5%, p = NS) or

0

10 000

20 000

30 000

40 000

50 000

60 000

70 000

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

Low frequency power


NV

n=22 n=23 n=20 n=20

mse

c

20

30

40

50

60

70

80

90 n=22 n=23 n=20 n=20


Total power

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

nu

0

10

20

30

40

50

60

Months

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

n=22 n=23 n=20 n=20

High frequency power

nu

0

2

4

6

8

10

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

Months


NV

Low to high frequency ratio

n=22 n=23 n=20 n=20

∗

∗

∗

∗

Fig. 2. Effect of tolrestat on frequency domain indices of heart rate variability in patients with diabetes mellitus (DM) and diabeticautonomic neuropathy (DAN) compared with placebo and controls. Values are expressed as medians with quartiles (Q1-25%,Q3-75%) plus minimum-maximum value ∗ p < 0.05 versus the 12th month of placebo. All frequency domain indices of both thecontrol groups were significantly higher than those of the two treatment groups. Abbreviations: nu = normalised units; NV = normalvolunteers.



on blood pressure was detected throughout thestudy.

Left Ventricular Systolic Function

At baseline, left ventricular ejection fractionwas similar in patients and controls (table I). Thiswas not changed by tolrestat or placebo adminis-tration and there were no differences in left ventric-ular ejection fraction values between patients andcontrols throughout the study.

Discussion

The present study demonstrated that the ARItolrestat slows down the progression of DAN, asevaluated by its effect on HRV level and standardcardiovascular reflex tests of autonomic function.In contrast, DAN deteriorated in placebo-treatedpatients over a 12-month period. At the twelfthtreatment month, HRV indices and the values ofsome cardiovascular reflex tests of autonomic

20

40

60

80

100

120

140

160

180

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

SDNN/24h

Severe DAN Moderate DAN

n=10 n=10 n=12 n=13

mse

c

0102030405060708090

100

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912


SDNN/5 min


n=10 n=10 n=12 n=13

0

4

8

12

16

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

pNN 50


n=10 n=10 n=12 n=13

RMSSD


n=10 n=10 n=12 n=13

0

10

20

30

40

50

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

mse

cm

sec

Months

Tolrestat Placebo Tolrestat Placebo

Months


∗

∗

∗∗

∗

∗∗

%

Fig. 3. Effect of tolrestat on time domain indices of heart rate variability in patients with diabetes mellitus (DM) and moderate(n = 12) or severe (n = 10) diabetic autonomic neuropathy (DAN) compared with placebo-treated patients with moderate (n = 13) orsevere (n = 10) diabetic autonomic neuropathy. Values are expressed as medians with quartiles (Q1-25%, Q3-75%) plus minimum-maximum value ∗ p < 0.05, and ∗∗ p < 0.01 versus the respective month of placebo-treated patients with analogous severity of DAN.Abbreviations: as in figure 1.



function of patients receiving tolrestat were signif-icantly better than those of patients on placebo.

The present study included only normotensivepatients with definite DAN free of coronary arterydisease, diabetic cardiomyopathy or heart failureof any aetiology, as we wanted to assess the neteffect of tolrestat on DAN, without interference ofleft ventricular systolic function impairment atbaseline or any potential drug-induced change.Patients receiving medication or having diseasesinterfering with HRV assessment were also ex-cluded.

In the tolrestat group, both time and frequencydomain HRV indices deteriorated at the third andsixth months of treatment in a mode similar to thatof the placebo group. At the ninth month, a partialrestoration of HRV indices was recorded and at thetwelfth month, some values, mainly those relatedto vagal tone (pNN50, RMSSD, HFP, andLFP/HFP ratio), were significantly better thanthose of patients on placebo. Compared with base-line, HRV indices were slightly, but not signifi-cantly, improved. Therefore, it is apparent thattolrestat requires a 12-month period to express its

0

4000

8000

12 000

16 000

20 000

24 000

28 000

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

Total power

Severe DAN Moderate DANn=10 n=10 n=12 n=13

∗

0102030405060708090

100110120

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912


Low frequency power


5

15

25

35

45

55

65

0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

High frequency power


Months


0 3 6 912 0 3 6 912 0 3 6 912 0 3 6 912

Months


Low to high frequency ratio


0

2

4

6

8

10

∗

mse

c

nu

nu

Fig. 4. Effect of tolrestat on frequency domain indices of heart rate variability in patients with diabetes mellitus (DM) and moderate(n = 12) or severe (n = 10) diabetic autonomic neuropathy (DAN) compared with placebo-treated patients with mod-erate (n = 13) or severe (n = 10) diabetic autonomic neuropathy. Values are expressed as medians with quartiles (Q1-25%, Q3-75%)plus minimum-maximum value ∗ p < 0.05 versus the respective month of placebo-treated patients with analogous severity of DAN.Abbreviations: as in figure 2.



effect on DAN. Nevertheless, no patient showeddeterioration in HRV indices.

On the other hand, in the placebo group mostpatients demonstrated a deterioration in all HRVindices, and some of these reached statistical sig-nificance (RMSSD, HFP, and LFP/HFP ratio).

There is no evidence that tolrestat can affectHRV level with mechanisms other than autonomicfunction improvement. Thus, it would appear rea-sonable to suggest that the partial restoration ofHRV indices towards ‘normal’ seen with tolrestatwas due to a direct beneficial effect of the drug onDAN. The initial and transient deterioration inHRV indices in the tolrestat group, similar to thatseen in patients on placebo, is indicative of the de-layed action of the drug. On the other hand,tolrestat ameliorated the progression of DANmainly in patients with moderate disease.

For the remaining patients with severe DAN, thepreservation of HRV indices to a level not signifi-cantly different to that of baseline may not beenough (in terms of absolute values) to protectthem from complex arrhythmias. This highlightsthe fact that early detection of DAN and initiationof treatment is essential for a clinically significantinterruption to DAN deterioration.

The difference in sympathetic-parasympatheticbalance between patients on placebo or tolrestat1 year after treatment initiation may contribute tothe reduction of risk for life-threatening ventriculararrhythmias in tolrestat-treated patients,[1-3]

mainly in those with moderate DAN. This is sup-ported by the significant reduction of prematureventricular beats in the tolrestat group, comparedwith placebo and baseline, which could be attrib-uted to a beneficial effect of the drug on sympatho-vagal interaction. This may contribute to the reduc-tion in the threshold of cardiac ectopic activity.

Previous studies with ARIs in DAN have pro-duced disappointing results. The reports regardingsorbinil efficacy have been conflicting,[19,20] andeventually the drug was withdrawn because ofincreased toxicity, including toxic epidermalnecrolysis and Stevens-Johnson syndrome.[21]

One study showed that ponalrestat amelioratedthe progression of DAN,[22] but two other studiesfailed to show any effect of this drug on DAN,probably because of the low doses adminis-tered.[23,24] Giugliano et al. showed that long-termtolrestat administration may contribute to the pri-mary prevention of DAN[25] and the improvementof mild DAN,[26] and this was confirmed by signi-ficant improvement of postural hypotension, E/Iratio and lying-to-standing (30:15) ratio in asymp-tomatic patients with mild DAN. However, thislatter study was criticised on the basis of whetherthe patients included did indeed have mild DAN orwhether they were borderline cases.[27] No otherstudy investigated the effect of an ARI on HRV inpatients with severe or moderate DAN, based on a24-hour electrocardiogram recording.

We believe that our findings are accurate be-cause: (a) the hard- and software used to assessHRV meet all requirements set by the InternationalTask Force on HRV,[1] (b) our laboratory has pre-vious experience with the use of method in post-myocardial infarction patients[16] and patients withDAN,[28] and finally (c) the use of two controlgroups provides protection from pitfalls. The effectof tolrestat on HRV indices is supported by thechanges in values in the cardiovascular autonomicfunction tests, some of which demonstrated a sig-nificant improvement, compared with placebo andbaseline.

Tolrestat has recently been withdrawn becauseof lack of effect on peripheral neuropathy, but newARIs are emerging and some of these are about tobe tested on humans. The evaluation of the effectof tolrestat on HRV indices and standard cardio-vascular autonomic function tests in patients withdefinite DAN in the present study may serve as areference for the assessment of the effect of newARIs on autonomic nervous system function.

Study Limitations

The relatively small number of patients andshort follow-up period did not allow the evaluationof the clinical implications of our findings on HRVlevel. Given that tolrestat has been withdrawn, the



assessment of the benefit induced by it on cardio-vascular morbidity and mortality could not be con-firmed by larger studies of longer duration. Hope-fully this can be accomplished with a new ARI inthe future.

Conclusion

Tolrestat ameliorates the progression of definiteDAN, mainly in its early stage, 12 months aftertreatment initiation, and significantly reduces pre-mature ventricular beats. This may contribute tothe reduction in risk for malignant ventriculararrhythmias and increased cardiovascular mortal-ity in these patients. The early detection of DAN isessential for effective intervention because nervedamage may still be reversible at that time.

AcknowledgementsWe would like to thank Dr Dimitrios Papanastasiou,

Associate Professor of Statistics, for his helpful guidance inthe statistical analysis of the rather complex data of thepresent study. No company or institution other than theHippocration Hospital has financed this study.

References1. Task Force of the European Society of Cardiology and the North

American Society of Pacing and Electophysiology. Heart ratevariability. Standards of measurement, physiological inter-pretation, and clinical use. Eur Heart J 1996; 17: 354-81

2. van Ravenswaaij CMA, Kollee LAA, Hopman JCW, et al.Heart rate variability. Ann Intern Med 1993; 118: 436-47

3. Genis AB, Guidino J, Vinolas X, et al. Autonomic influences ofventricular arrhythmias: Experimental and clinical aspects.In: Diez J, Dzau VJ, Ferrari R, et al., editors. Molecular cellbiology of cardiovascular diseases. Madrid: Mosby/DoymaLibros, 1995: 255-9

4. Regani M, Malfatto G, Pierini S, et al. Spectral analysis of heartrate variability in the assessment of autonomic diabetic neuro-pathy. J Auton Nerv Syst 1988; 23: 143-53

5. Ewing DJ, Campbell IW, Clarke BF. The natural history ofdiabetic autonomic neuropathy. Q J Med 1980; 193: 95-108

6. Sampson MJ, Wilson S, Karagiannis P, et al. Progression ofdiabetic autonomic neuropathy over a decade in insulin-dependent diabetics. Q J Med 1990; 278: 635-46

7. O’Brien IA, McFadden JP, Corrall RJM. The influence of dia-betic autonomic neuropathy on mortality in insulin-depend-ent diabetics. Q J Med 1991; 290: 495-502

8. Broomgarden ZT. Neuropathy and retinopathy. American Dia-betes Association Scientific Sessions, 1995. Diabetes Care1995; 18: 1314-7

9. The Diabetes Control and Complications Trial Research Group.The effect of intensive diabetes therapy on the developmentand progression of neuropathy. Ann Intern Med 1995; 122: 561-8

10. Ohkubo Y, Kishikawa H, Araki E, et al. Intensive insulintherapy prevents the progression of diabetic microvascular

complications in Japanese patients with non-insulin-depend-ent diabetes mellitus: a randomized prospective 6-year study.Diabetes Res Clin Pract 1995; 28: 103-17

11. Yagihashi S. Pathology and pathogenetic mechanisms of dia-betic neuropathy. Diabetes Metab Rev 1995; 11: 193-225

12. The American Diabetes Association and the American Acad-emy of Neurology. Consensus Statement. Standardized meas-ures in diabetic neuropathy. Diabetes Care 1995; 18 (1 Suppl.):59S-82S

13. Genovely H, Pfeifer MA. RR-variation: the autonomic test ofchoice in diabetes. Diabetes Metab Rev 1988; 3: 255-71

14. Boudoulas H, Mantzouratos D, Sohn YH, et al. Left ventricularmass and systolic performance in chronic systemic hyperten-sion. Am J Cardiol 1988; 57: 232-7

15. Kontopoulos A, Athyros V, Papageorgiou A, et al. Effect ofquinapril or metoprolol on heart rate variability in post-myo-cardial infarction patients. Am J Cardiol 1996; 77: 242-6

16. Jungsblut M. Erfassung and analyse der herzfrequenz-variabilitat auf basis des Langzeit-EKG-Systems. Thesis,Hamburg, Germany, 1993

17. Lombardi F, Sandrone G, Pernpruner S, et al. Heart rate vari-ability as an index of sympathovagal interaction after acutemyocardial infarction. Am J Cardiol 1987; 60: 1239-45

18. Spallone V, Uccioli L, Menzinger G. Diabetic autonomic neu-ropathy. Diabetes Metab Rev 1995; 11: 227-57

19. Fagius J, Brattberg A, Jameson S, et al. Limited benefit of treat-ment of diabetic polyneuropathy with an aldose reductaseinhibitor: a 24-week controlled trial. Diabetologia 1985; 28:323-9

20. Green A, Jaspan J, Kavin H, et al. Influence of long-term aldosereductase inhibitor therapy on autonomic dysfunction of uri-nary bladder, stomach and cardiovascular systems in diabeticpatients. Diabetes Res Clin Pract 1987; 4: 67-75

21. Pitts NE, Vreeland F, Shaw GL, et al. Clinical experience withsorbinil – an aldose reductase inhibitor. Metabolism 1986; 35(4 Suppl. 1): 96S-100S

22. Sundkvist G, Armstrong FM, Bradbury JE, et al. Peripheral andautonomic nerve function in 259 diabetic patients with peri-pheral neuropathy treated with ponalrestat (an aldose reduc-tase inhibitor) or placebo for 18 months. J DiabetesComplications 1992; 6: 123-30

23. Gill JS, Williams G, Ghatei MA, et al. Effect of the aldosereductase inhibitor, ponalrestat, on diabetic neuropathy.Diabéte Métab 1990; 16: 296-302

24. Ziegler D, Mayer P, Rathman W, et al. One-year treatment withaldose reductase inhibitor, ponalrestat, in diabetic neuro-pathy. Diabetes Res Clin Pract 1991; 14: 63-74

25. Giugliano D, Acampora R, Marfella R, et al. Tolrestat in theprimary prevention of diabetic neuropathy. Diabetes Care1995; 18: 536-41

26. Giugliano D, Marfella R, Quatraro A, et al. Tolrestat for milddiabetic neuropathy. A 52-week, randomized, placebo-con-trolled trial. Ann Intern Med 1993; 118: 7-11

27. Ziegler D. Diabetic cardiovascular neuropathy: prognosis, di-agnosis and treatment. Diabetes Metab Rev 1994; 10: 339-83

28. Kontopoulos AG, Athyros VG, Didangelos TP, et al. Effect ofchronic quinapril administration on heart rate variability inpatients with diabetic autonomic neuropathy. Diabetes Care1997; 20: 355-61

Correspondence and reprints: Dr Vasilios G. Athyros,15 Marmara St, Thessaloniki, 551 32, Greece.



Effect of Aldose Reductase Inhibition on Heart Rate Variability in Patients with Severe or Moderate...

Documents

Transcript of Effect of Aldose Reductase Inhibition on Heart Rate Variability in Patients with Severe or Moderate...