- Email: [email protected]
Why does diabetic autonomic neuropathy predict IDDM mortality? An analysis from the Pittsburgh epidemiology of diabetes complications study
ELSEVIER
Diabetes Research and Clinical Practice 34 Suppl. (1996) S165 SI71
Why does diabetic autonomic neuropathy predict IDDM mortality? An analysis from the Pittsburgh Epidemiology of Diabetes Complications Study T.J. Orchard ~''*, C.E. Lloyd ~', R.E. Maser b, L.H.
Kuller ~
"Department O~ El)M~'mio/oZ' Gradual~' &'lloo/ o/ Public llea/lh, U#urersity o/PiHsburg/z, Pitt.sburgll, P.4, USA I~J4eUico/ Tec/mo/og,~vPro
Abstract Back,,round: Previous studies have suggested that IDDM subjects with diabetic autonomic neuropathy (DAN) have a greatly increased risk of mortality which may relate to a specific cardiologic etiology. Ohjecth'es: To examine the predictors of DAN in 1DDM and its relationship to subsequent mortality. Study population: The Epidemiology of Diabetes Complications Study based on an incident cohort of childhood onset IDDM subjects. Data fi'om two examinations, separated by 2 years, are utilized. Methods: Diabetic autonomic neuropathy was determined by Expiration/Inspiration (E/I ratio). A variety of baseline risk factors were related to its subsequent incidence (n = 57 out of 325 subjects free of DAN at baseline). Two-year mortality by DAN status was also determined for all 479 subjects seen at baseline. Results: Duration of diabetes, the cardiovascular risk profile (hypertension, elevated LDL cholesterol and triglycerides), and other complications (e.g. nephropathy) were all univariately associated with subsequent DAN (P < 0.01). Smoking status and hemoglobin AI (HbA~), but less strongly, related (P < 0.05). Cox proportional hazards modeling showed diabetes duration and HbA~ to be significant independent predictors. Distal Symmetrical Polyneuropathy also contributed if added to the model. Mortality was increased four-fold in those with DAN ( P = 0.005), although this difference no longer was significant after adjustment for baseline nephropathy (P = 0.35) or hypertension (P = 0.42). Conclusions: Duration of diabetes and HbAI are the major predictors of DAN. However, although DAN is clearly associated with increased mortality, this is largely explained by associations with complications (e.g. nephropathy) and increased cardiovascular risk factors (e.g. hypertension). Kej,words: Diabetic autonomic neuropathy; Cardiovascular risk: Mortality; Glycemic control; Insulin dependent diabetes mellitus
1. Introduction In a l a n d m a r k study in 1976, Ewing r e p o r t e d that s y m p t o m a t i c diabetic a u t o n o m i c n e u r o p a t h y ( D A N ) [1] was associated with a high m o r t a l i t y . A 0168-822796,'S15.00 <~? 1996 Elsevier Science Ireland Ltd. All rights reserved PII SO168-8227(96)01321-6
S166
T,.I. Orchard el a/.
Diahele.~ Re.~earch am/ (7itHcal Praclice 34 Suppl. (1996) S165 S171
Table I Study population by age, sex, duration of IDDM EDC study, cycle I1
Age
Duration
Years
Men
Women
Total
< 18 18 29 30 < 10 10 14 15 19 20 24 25 29 30 -
13 86 78 3 41 45 41 27 20 177
11 72 65 3 37 37 26 24 21 148
24 158 143 6 78 82 67 51 41 325
Total
number of subsequent studies [2,3], including a further report by Ewing [4], have confirmed these observations. However, most of these reports have involved individuals with a long duration of IDDM and/or other concomitant complications. Consequently, the specific role of DAN is difficult to discern. Rathmann, in a more recent report, attempted to control for these influences by excluding subjects with advanced renal or cardiovascular disease and selecting a control group, which was carefully matched on age, sex and duration. These results showed a much smaller, but still significant, difference in 8-year mortality [5]. Though much of the excess mortality appears to be related to renal disease, some of the excess has been due to cardiovascular mortality, in particular, sudden cardiac death [1]. This has led to the concept that DAN may lead independently to sudden death. Another dimension to this possibility is the recognition that patients with diabetes may have an increased prevalence of asymptomatic myocardial ischemia [6]. However, though this has been demonstrated in many studies [7,8], a recent review of cardiology studies failed to demonstrate an increased prevalence of asymptomatic ischemia in diabetic patients compared to controls undergoing exercise stress tests
[9]. To further examine the association of diabetic autonomic neuropathy to mortality and cardiovascular risk, data from the Pittsburgh Epidemiology of Diabetes Complications Study were examined.
2. Materials and methods
The Epidemiology of Diabetes Complications Study (EDC) is a 10-year follow-up study of a cohort of insulin dependent diabetic patients identified from the Children's Hospital registry, who were diagnosed between 1950 and May 1980 at an age less than 17 years. This cohort has previously been shown to be representative of the Allegheny County population based registry with which it shares a 70% overlap [10]. Subjects were first seen between 1986 88 and are re-evaluated on a biennial basis. In Cycle II (1988 90), autonomic neuropathy testing was introduced for the entire cohort. Those attending Cycle I1 form the cohort of this report. The EDC study population and methods have been extensively described [11,12]. Diabetic autonomic neuropathy (DAN) was measured by means of the heart rate response to deep breathing. The test was performed after breakfast, approximately 1 h after injection of insulin, at the subject's biennial visit. A continuous electrocardiogram was performed during a l-rain breathing procedure which comprised six maximal expirations and inspirations. The E/I ratio was determined by the mean value of the longest RR interval during expiration and the shortest RR interval during inspiration. The procedure for the E/I ratio was repeated after a 1 rain break, and the mean E/I ratio of both trials was used for analysis. DAN was considered to be present if the mean E/I ratio was below 1.1. This procedure is similar to one previously described by Sundkvist [13].
T.J. Orchord ~': a/.
l)iahelc,~ Research am/ CliHical Praclicc 34 Sul~l#. (1996) S165 S171
SI67
Table 2 Prevalence (%) of autonomic symptom> b'r diabetic autonomic neuropathy ~ ( D A N ) 'status'* at follow-up Symptonl
DAN +
DAN
11 Postural hypotension Anorexia Gastric fullness Nocttlrnal diarrhea Incomplete bladder emptying Increased sweating Decreased hypoglycemic awarencs> Impotence (males only)
57 0.2 O.2 8.6 6.8 2.8 9.3 38.1 19.3
268 1.6 O.3 5.9 0.0 2.8 2.2 21.5 2.9
"Expiratiol~ h'lspifalion Ratio < I.I0. *All differences signilicant P < 0.0{11.
In addition to determination of DAN, distal symmetrical polyneuropathy (DSP) was determined using the DCCT clinical examination protocol [14] including systematic questioning for symptoms of autonomic neuropathy, e.g. orthostatic dizziness, early satiety, excessive diarrhea, severe constipation, genital/urinary symptoms, pseudomotor symptoms and reduced awareness of hypoglycemia. DSP was determined by the presence of at least two of the three following states: symptoms and/or signs consistent with DSP and absent (or decreased) tendon reflexes, all in the absence of other known etiologies. Retinopathy was determined on the basis of stereo fundus photographs read by the Wisconsin Reading Center and graded according to the Modified Airlie House System [15]. Nephropathy was determined on the basis of the albumin excretion rate (AER) from at least two of three urine samples (i.e., 24 h. overnight and 4 h clinic sample) being positive. The criteria for microalbuminuria was AER 20 200 /;g/min, and for overt nephropathy, > 200 itg/min. Albumin was measured immunonephlemetrically [16]. Coronary heart disease was determined by the presence of physician-diagnosed angina during clinic visit and/or contirmed angiography with 50 + % coronary artery stenosis a n d o r a history of myocardial infarction meeting standard criteria [17]. A number of risk factors were also determined including, on a fasting blood draw, a complete lipoprotein profile. Total cholesterol [18] and
triglycerides [19] were measured enzymatically, and high density lipoprotein cholesterol determilled after manganese precipitation by modification [20] of the lipid research clinic's methods [21]. Low density lipoprotein cholesterol levels were calculated by the Friedewald formula [22]. Fibrinogen levels were performed with a buret calorimetric procedure and a clotting method. Stable glycosylated hemoglobin was measured using an automated high performance liquid chromatography method (DIAMAT. Bio-Rad Laboratories, Hercules, CA). Blood pressure was measured with a random zero sphygmomanometer after 5 rain quite rest m a seated position [23]. Three readings were taken, and the mean of the last two was used in analysis. Hypertension was detined as a blood pressure above 140:90 and:or being on antihypertensive medication. % Prevalence 80 60
40
j_.~
20 0
5-9
10-14
%__-~_J--y 15-19
20-24
25-29
30+
Duration (years) Prevalence of DAN (E/I <1.10): (,,) and of 2 or more Autonomic Symptoms: to). Cycle III Fig. 1. Prevalence of D A N (E 1 < 1.10} and of two or more autonomic symptoms. Cycle 11I.
S168
T.J. Orchard et at.
Diahet~'.s Research am/ CIh~ica/ Practice 34 Suppl. (1996) S165 S171
Table 3 Univariate baseline (Cycle Ill predictors ('i, m e a n - S.D.) of the 2-year incidence (Cycle I11) of DAN~: EDC study
n (% male) Duration (years) Ever smoker (',) Hypertension (%) HbA1 (%) L D L cholesterol ring dl/ H D L cholesterol (rag dl) Triglycerides (nag dl) Fibrinogen (rag dl) Insulin dose (per kg) Nephropathy (;';1) Proliferative retinopathy (%) Distal symmetrical polyneuropathy (':) Cardiovascular disease (%)
Incident DAN
No DAN
P-values
57 {47) 25.2 + 8.3 44 33 I1.1 +2.0 127.0 + 35.9 53.8 + 13.7 121.2 ± 68.2 318.7 _+ 96.6 0.71 + 0.21 26 53 40 9
268 (56) 19.8 _+ 6.6 24 11 10.5_+ 1.7 109.0 + 33.1 53.4 + 12.4 94.7 + 69.5 277.5 - 88.3 0.80 + 0.2 13 23 11 2
0.236 < 0.0l)1 0.041 < 0.0l)1 0.014 < 0.001 0.804 0.900 0.002 0.010 0.009 < 0.(101 < 0.001 0.013
"E:I ratio <1.10.
For statistical analysis, means and distributions were compared between those with and without incident DAN by Student's t-test (with log transformation of variables, e.g. triglycerides when needed) and ze-testing as appropriate. The 2-year incidence of DAN was also examined using Cox proportional hazards modeling. Relative risk for mortality by DAN status were calculated both unadjusted and after independent adjustment for duration, coronary heart disease, nephropathy or hypertension by the Mantel-Haensel test.
3. Results
Table 1 shows the study population for this report. A total of 325 subjects had full clinical examination at Cycle 1I and Cycle III. and did not have evidence of DAN at Cycle 1I and thus were eligible for this incidence analysis. As can be seen in Table 2, the prevalence of autonomic symptoms (at Cycle III) is uniformly and very significantly increased in those with an abnormal E/I ratio compared to those without. This increase range from a 34-fold increase (anorexia) to a 1.8-fold increase li)r hypoglycemic unawareness. A strong association between these two manifestations of DAN is shown in Fig. 1, where the prevalence of DAN (diagnosed by E/1 ratio <
1.1 ) is compared to the prevalence of two or more autonomic symptoms. A very close correspondence between the two is apparent. Approximately 70% of IDDM patients, after 30 or more years of IDDM, have DAN. The rest of the analysis focuses on the incidence and mortality associated with DAN as defined by E/I ratio alone. Table 3 shows that a number of univariate differences at Cycle 11 were noted between those 57 who subsequently developed DAN over the next 2 years and the 268 subjects who did not. Duration, hemoglobin A l, and the percentage who had ever smoked were all increased, as were the prevalence of other complications including DSP and the cardiovascular risk factors of hypertension, LDL cholesterol, fibrinogen and triglycerides. The incidence of DAN did not vary by either H D L cholesterol or gender. Forward stepwise Cox proportional hazards modeling (Table 4) showed duration (Model ~ l) to be the most significant predictor (P < 0.001) followed by hemoglobin A1 (P < 0.003) and a borderline contribution of fibrinogen ( = 0.05). If DSP was added to the model (Model ~ 2), it also contributed ( P < 0 . 0 0 1 ) and reduced, but did not eliminate, the relative risks for duration and hemoglobin A1. Two-year mortality was determined for the 88 subjects with DAN at Cycle 1I, and the 399
T.J. Orcllard el al.
Dfllhele,~ Rusearclt omt Clhtica/ Practice 34 Supt)/. (1996) S I 6 5 S171
S169
Table 4 Cox proportional hazard model: 2-year incidence ot DAN Model =4-1
Model ~-2
Variable
C o e f f S.E.
P-value
Variable
Coeff: S. E.
P-value
Duration HbA 1 Fibrinogcn
4.48 2.81 2.06
0.000 0.003 0.051
DSP Duration HbA 1 Fibrinogen
2.79 3.07 2.76 1.98
0.000 0.009 0.005 0.060
DSP, Distal symmetrical polyneuropathy.
subjects without DAN. Mortality was significantly higher in those with DAN (9.1%) compared to those without (2.3%), a Relative Risk of 4.03 (95% C . 1 . - 1 . 6 0 10.15). Cause of death as stated on the death certificate did not differ between the two groups; four of the eight with DAN died fl'om coronary heart disease, as opposed to five of the nine without. There was one renal failure case in those with DAN, one accident, one diabetic coma and one anoxic encephalopathy. The remaining mortality in those without DAN was related to infection (1), diabetic coma (2) and hypertrophic cardiomyopathy (1). Table 5 shows the relative risks for mortality by DAN status, both unadjusted and adjusted. As can be seen, the significant unadjusted relative risk was marginally reduced after adjustment for coronary heart disease or duration. However, when either nephropathy or hypertension is accounted for, it is considerably reduced and no longer significant.
4. Discussion
These data suggest a relatively high incidence of DAN (approximately 9% per year) which is considerably higher than the incidence of other complications, e.g. 4% per year for proliferative retinopathy or 3.5% for microalbuminuria [24]. The major risk factors, duration and HbA1, are also consistent with those seen for DSP [25]. These results confirm, albeit in a short follow-up period, an increased risk of mortality associated with DAN identified by abnormal E/I ratio. The relative risk is comparable to most previous studies (e.g. 5-fold in O'Brien's [2] study) though
lower than in Rathman's [5] report (8-fold). These data do not suggest a major excess of cardiovascular deaths (or sudden deaths) in this group. However, they do suggest that much of the association with mortality can be explained by the high prevalence of nephropathy in those with DAN~ and the consequential increased cardiovascular risk experienced by these subjects. It is well recognized that subjects with 1DDM who have renal disease have an increased risk of cardiovascular disease [26] and it has recently been argued that this may largely reflect the disturbed cardiovascular risk profile seen in patients with diabetic nephropathy [27]. It is important to note that in Ewing's original study [1], six of the 10 deaths were due to renal failure, and O'Brien et al. [2] also reported a 6-fold increase in the prevalence of proteinuria in those with DAN. DAN and nephropathy are thus closely linked. In an earlier study [28] we demonstrated, on a subgroup (aged 25 34) seen at Cycle 1 that were given a full neurological evaluation, similar crosssectional associations. In this case, hypertension, low and high density lipoprotein cholesterol concentrations were independent correlates of DAN. These results suggest that the disturbed CVD risk profile seen in nephropathy may lead to both CVD and autonomic neuropathy. Whether it is necessary to evoke a specific DAN mechanism leading to sudden arrhythmic death to explain this linkage seems unlikely although some association between DAN, a prolonged QT interval and sudden death has been suggested [29]. Further followup will enable us to distinguish the exact roles of cardiovascular risk factors, nephropathy and DAN in the etiology of cardiovascular disease
SI70
T.J. Orchard ('t al.
Dlahvtc~ Rc,~vacch and Clinical Practice, 34 SupFI. (1996) SI65 S171
Table 5 Relatixe risks for 2-)ear morlalil, b) D A N ' status ED(" study: ad ustment for duration, coronary hearl disease, nephropathy, hypertension Adjustment
Relati;e risks
95'!, C.I.
P-wduc
Nolle Coronar', heart disease Duration Nephropathy Hypertension
4.03 3.45 2.11 1.40 1.37
1.60 2.14 1.25 0.85 0.80
0.005 0.032 0.06 0.35 I).42
-
10,15 5.57 3.55 2.31 2.32
"E 1 ratio < 1.10.
(particularly as Holter momtoring l\)r myocardial ischemia and arrythmia is now being performed). However. it seems likely that, at this lime, much of the association is explained by a shared cardiovascular/renal risk, and thal prevention of the excess mortality associated with DAN may be best addressed by the prevention of renal and cardiovascular disease in IDDM. The well-known benefits of glycemic control [30], ACE inhibitors [31] and possibly (though currently not evaluated), lipid modulation should be encouraged. Acknowledgements This study was supported by National Institutes of Health, Grant No. DK 34818. We would like to acknowledge the assistance of Dr. Dorothy Becker (HbAI measures) and Dr. Demetrius Ellis (urinary albumin measures) and the willing participants in this study. References
[I] Ewing. D.J.. Campbelh I.W. and Clarke, B.F. (1976) Mortality In diabetic atttonomic neuropathy. Lancet 601 603. [2] O'Brien, I.A., McFadden. ,l.P. and Corralh R.J.M. (1991) The influence of atutonomic neuropathy on mortality in insulin-dependent diabetes. Q..I. Med. 79(290), 495 502. [3] Sampson, M.J., Wilson, S.. Karagiannis, P., Edmonds, M. and Watkins P. (1990) Progression of diabetic autonomic neuropathy over at decadc in insulin-dependent diabetics. Q. J. Med. 178, 635 645. [4] Ewing, D.J., Campbell. l.W. and Clarke, B.F. (1980) The natural history of diabetic autonomic neuropathy. Q. J. Med. 193, 95 108.
15] Rathmaun, W.. Ziegler, D., Jahnkc, M., ltaastert, B. and Gries, V.A. (19931 Mortality in diabetic patients with cardiovascular autonomic neuropath~. Diabetic Med. 10, 820 824. [6] Ncsto. R.W., Phillips. R.T., Kett. S.G, et all. (1988) Angina aud exertional myocardial ischemia in diabetic and nondiabetic patients: assessment by exercise thallium scintigraphy. Ann. Intern. Med. 108, 170 175. [7] Margolis, J.R.. Kannell, W.S., Feinleig, M., Dawber, T.R. and McNamara, P.M. (1973) Clinical features of tmrecognized myocardial infarction silent and symptomatic. Eighteen year follow-up: Ttle Framingham Study. Am. J. Cardiol. 32. 1 7. [8] Chiariello, M., lndolfi, C., Cotecchia. M.R., Sifola, C.. Romano, M. and Condorelli, M. (1985) Asymptomatic transient ST changes during ambulatory ECG monitoring m diabetic patients. Am. Heart J. 110, 529 534. [9] Juhanti Airaksinen, K.E. and Juhani Koislinen, M. (1992) Association between silent coronary artery disease, diabetes and autonomic neuropathy. Diabetes Care 15(2), 288 292. [10] Wagener, D.K., Soaks, J.M., LaPorte, R.E. and MacGregor, J.M. (1982) The Pittsburgh study of insulin-dependent diabetes mellitus: risk for diabetes among relatives of IDDM. Diabetes 31, 136 144. [11] Orchard, T.J., Dormam J.S., Maser, R.E. et al. (1990) Factors associated with the avoidance of severe complicalions after 25 years of insulin-dependent diabetes mellitus: Pittsburgh Epidemiology of Diabetes Complications Study-1. Diabetes ('are 13, 741 747. [12] Orchard, T.J., Dorman, J.S., Maser. R.E. et al. (1990) The prevalence of complications in insulin-dependent diabetes mellitus by sex and duration: Pittsburgh Epidemiology of Diabetes Complications Study-ll. Diabetes 39. 1116 1124. [13] Sundkvist. G. (1981) Autonomic nervous function m asymptomatic diabetic patients with signs of peripheral neuropathy. Diabetes Care 4, 529 534. [14] "['he Diabetes Control and Complications Trial Research Group {1987) Manual of Operations Ibr the Diabetes Control and Complications Trial. Washington, DC: US Dept of Commerce.
T.J. Or
DiaDc:c,~ Research am/ C/iHi~al Prac:icc 34 Suppl. (1996) S I 6 5 S171
[15] Early Treatment Diabetic Retinopalhy Study Coordinating Center (1980) Manual of Operations. Baltimore, M D: University of Maryland School of Medicine, Diabetic Retinopathy Coordinating Center: Chapters 5 and 18. [16] Ellis, D. and Becket, D.J. (1974) Editorial reply. J. Pediatr. 104, 324 325. [I 7] Orchard, T.J. on behalf of the Community Cardiowtscular Surveillance Project investigators (1985) Validation of coronary heart disease mortality data: the Community Cardiovascular Surveillance Project Pilot Experience. CVD Epid. Newslett. 157, 46. [18] Allain, C., Pooll, LS.. Chan, C.S.(}., Richmond, W. and Fu, P.C. {1974) Enzymatic determination of total serum cholesteroh Clin. Chem. 20, 470 475. [19] Bucolo, G. and David, H.(1973)Quantitativedctermination of serum triglyceridcs by usc of cnzylnes. Clin. Chem. 19, 476 482. [20] Warnick, G.R. and Albcrs, J.J. (1978) tteparin Mn -~ quantitation of high density lipoprotein cholesterol: an ultrafiltralion procedure l'or lipcmic samples. Clin. Chem. 24, 9{){) 904. [21] National Institutes of Heahh. Department of ttealth, Education and Welfare (19751 Lipid Research Clinics Program. Washington, DC: Government Printing Office: 9-50. Publication No. Nltt 75-628. [22] Fricdewald, W.T., Levy, R.I. and Ercdrickson, D.S. (1972) Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin. Chem. 18, 499 502. [23] Borhani, N.O., Kass. f-.H., l,angford, H.G.. Payne, G.H., Remington, R.D. and Stamler. J. (1976) The hypertension detection and follow-up program. Prey. Med. 5. 207 215. [24] Lloyd, C.E., Becker. D.. Ellis. D. and Orchard. T.J.
[25]
[26]
[27]
[28]
[29]
[3(I]
[31]
SI7I
(1996) Incidence of complications in insulin-dependent diabetes mellitus: A survival analysis. Am. J. Epidemiol. 143. 431 441, Wilson, R, and Orchard. T.J. (1994) llypertension is a major risk factor for distal symmetrical polyneuropathy (DSP) in insulin dependent diabetes mellitus. American Diabetes Association, New Orleans, Diabetes 43(I), =4-42. Borch-Johnsen, K., Andersen, P.K. and Deckert, T. (1985) The effect of proteinuria on relative mortality in type I (insulin-dependent) diabetes mellitus. Diabetologia 28, 590 596. Donahue, R.P. and Orchard, T.J. (1992) Diabetes mellitus and macrovascular complications. An epidemiological perspective. Diabetes Care 15(9), 1141 1155. Maser, R.E.. Pfeilcr, M.A., Dorman, J.S.. KuNer, L.H., Becker, D.J. and Orchard, T.J. (1990) Diabetic autonomic neuropathy and cardiovascular risk. Pittsburgh Epidemiology of Diabetes Complications Study 111. Arch. lntern. Med. 150, 1218 1222. Ewing, D.J., Boland. O., Neilson, J.M.M., Cho, C.G. and Clarke. B.F. (1991) Autonomic neuropathy, QT interval lengthening and unexpected deaths in male diabetic patients. Diabelologia 34. 182 185. The Diabetes Control and Complications Trial Research Group (1993) The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N. Engh ,I. Med. 329, 977 986. Lewis, E.J., Hunsicker, L.G., Bain, R.P. and Rohde, R.D. (1993) The effect of angiotensin-con\,erting-enzyme inhibition on diabetic nephropathy. N. Engh J. Med. 329, 1456 1462.
Diabetes Research and Clinical Practice 34 Suppl. (1996) S165 SI71
Why does diabetic autonomic neuropathy predict IDDM mortality? An analysis from the Pittsburgh Epidemiology of Diabetes Complications Study T.J. Orchard ~''*, C.E. Lloyd ~', R.E. Maser b, L.H.
Kuller ~
"Department O~ El)M~'mio/oZ' Gradual~' &'lloo/ o/ Public llea/lh, U#urersity o/PiHsburg/z, Pitt.sburgll, P.4, USA I~J4eUico/ Tec/mo/og,~vPro
Abstract Back,,round: Previous studies have suggested that IDDM subjects with diabetic autonomic neuropathy (DAN) have a greatly increased risk of mortality which may relate to a specific cardiologic etiology. Ohjecth'es: To examine the predictors of DAN in 1DDM and its relationship to subsequent mortality. Study population: The Epidemiology of Diabetes Complications Study based on an incident cohort of childhood onset IDDM subjects. Data fi'om two examinations, separated by 2 years, are utilized. Methods: Diabetic autonomic neuropathy was determined by Expiration/Inspiration (E/I ratio). A variety of baseline risk factors were related to its subsequent incidence (n = 57 out of 325 subjects free of DAN at baseline). Two-year mortality by DAN status was also determined for all 479 subjects seen at baseline. Results: Duration of diabetes, the cardiovascular risk profile (hypertension, elevated LDL cholesterol and triglycerides), and other complications (e.g. nephropathy) were all univariately associated with subsequent DAN (P < 0.01). Smoking status and hemoglobin AI (HbA~), but less strongly, related (P < 0.05). Cox proportional hazards modeling showed diabetes duration and HbA~ to be significant independent predictors. Distal Symmetrical Polyneuropathy also contributed if added to the model. Mortality was increased four-fold in those with DAN ( P = 0.005), although this difference no longer was significant after adjustment for baseline nephropathy (P = 0.35) or hypertension (P = 0.42). Conclusions: Duration of diabetes and HbAI are the major predictors of DAN. However, although DAN is clearly associated with increased mortality, this is largely explained by associations with complications (e.g. nephropathy) and increased cardiovascular risk factors (e.g. hypertension). Kej,words: Diabetic autonomic neuropathy; Cardiovascular risk: Mortality; Glycemic control; Insulin dependent diabetes mellitus
1. Introduction In a l a n d m a r k study in 1976, Ewing r e p o r t e d that s y m p t o m a t i c diabetic a u t o n o m i c n e u r o p a t h y ( D A N ) [1] was associated with a high m o r t a l i t y . A 0168-822796,'S15.00 <~? 1996 Elsevier Science Ireland Ltd. All rights reserved PII SO168-8227(96)01321-6
S166
T,.I. Orchard el a/.
Diahele.~ Re.~earch am/ (7itHcal Praclice 34 Suppl. (1996) S165 S171
Table I Study population by age, sex, duration of IDDM EDC study, cycle I1
Age
Duration
Years
Men
Women
Total
< 18 18 29 30 < 10 10 14 15 19 20 24 25 29 30 -
13 86 78 3 41 45 41 27 20 177
11 72 65 3 37 37 26 24 21 148
24 158 143 6 78 82 67 51 41 325
Total
number of subsequent studies [2,3], including a further report by Ewing [4], have confirmed these observations. However, most of these reports have involved individuals with a long duration of IDDM and/or other concomitant complications. Consequently, the specific role of DAN is difficult to discern. Rathmann, in a more recent report, attempted to control for these influences by excluding subjects with advanced renal or cardiovascular disease and selecting a control group, which was carefully matched on age, sex and duration. These results showed a much smaller, but still significant, difference in 8-year mortality [5]. Though much of the excess mortality appears to be related to renal disease, some of the excess has been due to cardiovascular mortality, in particular, sudden cardiac death [1]. This has led to the concept that DAN may lead independently to sudden death. Another dimension to this possibility is the recognition that patients with diabetes may have an increased prevalence of asymptomatic myocardial ischemia [6]. However, though this has been demonstrated in many studies [7,8], a recent review of cardiology studies failed to demonstrate an increased prevalence of asymptomatic ischemia in diabetic patients compared to controls undergoing exercise stress tests
[9]. To further examine the association of diabetic autonomic neuropathy to mortality and cardiovascular risk, data from the Pittsburgh Epidemiology of Diabetes Complications Study were examined.
2. Materials and methods
The Epidemiology of Diabetes Complications Study (EDC) is a 10-year follow-up study of a cohort of insulin dependent diabetic patients identified from the Children's Hospital registry, who were diagnosed between 1950 and May 1980 at an age less than 17 years. This cohort has previously been shown to be representative of the Allegheny County population based registry with which it shares a 70% overlap [10]. Subjects were first seen between 1986 88 and are re-evaluated on a biennial basis. In Cycle II (1988 90), autonomic neuropathy testing was introduced for the entire cohort. Those attending Cycle I1 form the cohort of this report. The EDC study population and methods have been extensively described [11,12]. Diabetic autonomic neuropathy (DAN) was measured by means of the heart rate response to deep breathing. The test was performed after breakfast, approximately 1 h after injection of insulin, at the subject's biennial visit. A continuous electrocardiogram was performed during a l-rain breathing procedure which comprised six maximal expirations and inspirations. The E/I ratio was determined by the mean value of the longest RR interval during expiration and the shortest RR interval during inspiration. The procedure for the E/I ratio was repeated after a 1 rain break, and the mean E/I ratio of both trials was used for analysis. DAN was considered to be present if the mean E/I ratio was below 1.1. This procedure is similar to one previously described by Sundkvist [13].
T.J. Orchord ~': a/.
l)iahelc,~ Research am/ CliHical Praclicc 34 Sul~l#. (1996) S165 S171
SI67
Table 2 Prevalence (%) of autonomic symptom> b'r diabetic autonomic neuropathy ~ ( D A N ) 'status'* at follow-up Symptonl
DAN +
DAN
11 Postural hypotension Anorexia Gastric fullness Nocttlrnal diarrhea Incomplete bladder emptying Increased sweating Decreased hypoglycemic awarencs> Impotence (males only)
57 0.2 O.2 8.6 6.8 2.8 9.3 38.1 19.3
268 1.6 O.3 5.9 0.0 2.8 2.2 21.5 2.9
"Expiratiol~ h'lspifalion Ratio < I.I0. *All differences signilicant P < 0.0{11.
In addition to determination of DAN, distal symmetrical polyneuropathy (DSP) was determined using the DCCT clinical examination protocol [14] including systematic questioning for symptoms of autonomic neuropathy, e.g. orthostatic dizziness, early satiety, excessive diarrhea, severe constipation, genital/urinary symptoms, pseudomotor symptoms and reduced awareness of hypoglycemia. DSP was determined by the presence of at least two of the three following states: symptoms and/or signs consistent with DSP and absent (or decreased) tendon reflexes, all in the absence of other known etiologies. Retinopathy was determined on the basis of stereo fundus photographs read by the Wisconsin Reading Center and graded according to the Modified Airlie House System [15]. Nephropathy was determined on the basis of the albumin excretion rate (AER) from at least two of three urine samples (i.e., 24 h. overnight and 4 h clinic sample) being positive. The criteria for microalbuminuria was AER 20 200 /;g/min, and for overt nephropathy, > 200 itg/min. Albumin was measured immunonephlemetrically [16]. Coronary heart disease was determined by the presence of physician-diagnosed angina during clinic visit and/or contirmed angiography with 50 + % coronary artery stenosis a n d o r a history of myocardial infarction meeting standard criteria [17]. A number of risk factors were also determined including, on a fasting blood draw, a complete lipoprotein profile. Total cholesterol [18] and
triglycerides [19] were measured enzymatically, and high density lipoprotein cholesterol determilled after manganese precipitation by modification [20] of the lipid research clinic's methods [21]. Low density lipoprotein cholesterol levels were calculated by the Friedewald formula [22]. Fibrinogen levels were performed with a buret calorimetric procedure and a clotting method. Stable glycosylated hemoglobin was measured using an automated high performance liquid chromatography method (DIAMAT. Bio-Rad Laboratories, Hercules, CA). Blood pressure was measured with a random zero sphygmomanometer after 5 rain quite rest m a seated position [23]. Three readings were taken, and the mean of the last two was used in analysis. Hypertension was detined as a blood pressure above 140:90 and:or being on antihypertensive medication. % Prevalence 80 60
40
j_.~
20 0
5-9
10-14
%__-~_J--y 15-19
20-24
25-29
30+
Duration (years) Prevalence of DAN (E/I <1.10): (,,) and of 2 or more Autonomic Symptoms: to). Cycle III Fig. 1. Prevalence of D A N (E 1 < 1.10} and of two or more autonomic symptoms. Cycle 11I.
S168
T.J. Orchard et at.
Diahet~'.s Research am/ CIh~ica/ Practice 34 Suppl. (1996) S165 S171
Table 3 Univariate baseline (Cycle Ill predictors ('i, m e a n - S.D.) of the 2-year incidence (Cycle I11) of DAN~: EDC study
n (% male) Duration (years) Ever smoker (',) Hypertension (%) HbA1 (%) L D L cholesterol ring dl/ H D L cholesterol (rag dl) Triglycerides (nag dl) Fibrinogen (rag dl) Insulin dose (per kg) Nephropathy (;';1) Proliferative retinopathy (%) Distal symmetrical polyneuropathy (':) Cardiovascular disease (%)
Incident DAN
No DAN
P-values
57 {47) 25.2 + 8.3 44 33 I1.1 +2.0 127.0 + 35.9 53.8 + 13.7 121.2 ± 68.2 318.7 _+ 96.6 0.71 + 0.21 26 53 40 9
268 (56) 19.8 _+ 6.6 24 11 10.5_+ 1.7 109.0 + 33.1 53.4 + 12.4 94.7 + 69.5 277.5 - 88.3 0.80 + 0.2 13 23 11 2
0.236 < 0.0l)1 0.041 < 0.0l)1 0.014 < 0.001 0.804 0.900 0.002 0.010 0.009 < 0.(101 < 0.001 0.013
"E:I ratio <1.10.
For statistical analysis, means and distributions were compared between those with and without incident DAN by Student's t-test (with log transformation of variables, e.g. triglycerides when needed) and ze-testing as appropriate. The 2-year incidence of DAN was also examined using Cox proportional hazards modeling. Relative risk for mortality by DAN status were calculated both unadjusted and after independent adjustment for duration, coronary heart disease, nephropathy or hypertension by the Mantel-Haensel test.
3. Results
Table 1 shows the study population for this report. A total of 325 subjects had full clinical examination at Cycle 1I and Cycle III. and did not have evidence of DAN at Cycle 1I and thus were eligible for this incidence analysis. As can be seen in Table 2, the prevalence of autonomic symptoms (at Cycle III) is uniformly and very significantly increased in those with an abnormal E/I ratio compared to those without. This increase range from a 34-fold increase (anorexia) to a 1.8-fold increase li)r hypoglycemic unawareness. A strong association between these two manifestations of DAN is shown in Fig. 1, where the prevalence of DAN (diagnosed by E/1 ratio <
1.1 ) is compared to the prevalence of two or more autonomic symptoms. A very close correspondence between the two is apparent. Approximately 70% of IDDM patients, after 30 or more years of IDDM, have DAN. The rest of the analysis focuses on the incidence and mortality associated with DAN as defined by E/I ratio alone. Table 3 shows that a number of univariate differences at Cycle 11 were noted between those 57 who subsequently developed DAN over the next 2 years and the 268 subjects who did not. Duration, hemoglobin A l, and the percentage who had ever smoked were all increased, as were the prevalence of other complications including DSP and the cardiovascular risk factors of hypertension, LDL cholesterol, fibrinogen and triglycerides. The incidence of DAN did not vary by either H D L cholesterol or gender. Forward stepwise Cox proportional hazards modeling (Table 4) showed duration (Model ~ l) to be the most significant predictor (P < 0.001) followed by hemoglobin A1 (P < 0.003) and a borderline contribution of fibrinogen ( = 0.05). If DSP was added to the model (Model ~ 2), it also contributed ( P < 0 . 0 0 1 ) and reduced, but did not eliminate, the relative risks for duration and hemoglobin A1. Two-year mortality was determined for the 88 subjects with DAN at Cycle 1I, and the 399
T.J. Orcllard el al.
Dfllhele,~ Rusearclt omt Clhtica/ Practice 34 Supt)/. (1996) S I 6 5 S171
S169
Table 4 Cox proportional hazard model: 2-year incidence ot DAN Model =4-1
Model ~-2
Variable
C o e f f S.E.
P-value
Variable
Coeff: S. E.
P-value
Duration HbA 1 Fibrinogcn
4.48 2.81 2.06
0.000 0.003 0.051
DSP Duration HbA 1 Fibrinogen
2.79 3.07 2.76 1.98
0.000 0.009 0.005 0.060
DSP, Distal symmetrical polyneuropathy.
subjects without DAN. Mortality was significantly higher in those with DAN (9.1%) compared to those without (2.3%), a Relative Risk of 4.03 (95% C . 1 . - 1 . 6 0 10.15). Cause of death as stated on the death certificate did not differ between the two groups; four of the eight with DAN died fl'om coronary heart disease, as opposed to five of the nine without. There was one renal failure case in those with DAN, one accident, one diabetic coma and one anoxic encephalopathy. The remaining mortality in those without DAN was related to infection (1), diabetic coma (2) and hypertrophic cardiomyopathy (1). Table 5 shows the relative risks for mortality by DAN status, both unadjusted and adjusted. As can be seen, the significant unadjusted relative risk was marginally reduced after adjustment for coronary heart disease or duration. However, when either nephropathy or hypertension is accounted for, it is considerably reduced and no longer significant.
4. Discussion
These data suggest a relatively high incidence of DAN (approximately 9% per year) which is considerably higher than the incidence of other complications, e.g. 4% per year for proliferative retinopathy or 3.5% for microalbuminuria [24]. The major risk factors, duration and HbA1, are also consistent with those seen for DSP [25]. These results confirm, albeit in a short follow-up period, an increased risk of mortality associated with DAN identified by abnormal E/I ratio. The relative risk is comparable to most previous studies (e.g. 5-fold in O'Brien's [2] study) though
lower than in Rathman's [5] report (8-fold). These data do not suggest a major excess of cardiovascular deaths (or sudden deaths) in this group. However, they do suggest that much of the association with mortality can be explained by the high prevalence of nephropathy in those with DAN~ and the consequential increased cardiovascular risk experienced by these subjects. It is well recognized that subjects with 1DDM who have renal disease have an increased risk of cardiovascular disease [26] and it has recently been argued that this may largely reflect the disturbed cardiovascular risk profile seen in patients with diabetic nephropathy [27]. It is important to note that in Ewing's original study [1], six of the 10 deaths were due to renal failure, and O'Brien et al. [2] also reported a 6-fold increase in the prevalence of proteinuria in those with DAN. DAN and nephropathy are thus closely linked. In an earlier study [28] we demonstrated, on a subgroup (aged 25 34) seen at Cycle 1 that were given a full neurological evaluation, similar crosssectional associations. In this case, hypertension, low and high density lipoprotein cholesterol concentrations were independent correlates of DAN. These results suggest that the disturbed CVD risk profile seen in nephropathy may lead to both CVD and autonomic neuropathy. Whether it is necessary to evoke a specific DAN mechanism leading to sudden arrhythmic death to explain this linkage seems unlikely although some association between DAN, a prolonged QT interval and sudden death has been suggested [29]. Further followup will enable us to distinguish the exact roles of cardiovascular risk factors, nephropathy and DAN in the etiology of cardiovascular disease
SI70
T.J. Orchard ('t al.
Dlahvtc~ Rc,~vacch and Clinical Practice, 34 SupFI. (1996) SI65 S171
Table 5 Relatixe risks for 2-)ear morlalil, b) D A N ' status ED(" study: ad ustment for duration, coronary hearl disease, nephropathy, hypertension Adjustment
Relati;e risks
95'!, C.I.
P-wduc
Nolle Coronar', heart disease Duration Nephropathy Hypertension
4.03 3.45 2.11 1.40 1.37
1.60 2.14 1.25 0.85 0.80
0.005 0.032 0.06 0.35 I).42
-
10,15 5.57 3.55 2.31 2.32
"E 1 ratio < 1.10.
(particularly as Holter momtoring l\)r myocardial ischemia and arrythmia is now being performed). However. it seems likely that, at this lime, much of the association is explained by a shared cardiovascular/renal risk, and thal prevention of the excess mortality associated with DAN may be best addressed by the prevention of renal and cardiovascular disease in IDDM. The well-known benefits of glycemic control [30], ACE inhibitors [31] and possibly (though currently not evaluated), lipid modulation should be encouraged. Acknowledgements This study was supported by National Institutes of Health, Grant No. DK 34818. We would like to acknowledge the assistance of Dr. Dorothy Becker (HbAI measures) and Dr. Demetrius Ellis (urinary albumin measures) and the willing participants in this study. References
[I] Ewing. D.J.. Campbelh I.W. and Clarke, B.F. (1976) Mortality In diabetic atttonomic neuropathy. Lancet 601 603. [2] O'Brien, I.A., McFadden. ,l.P. and Corralh R.J.M. (1991) The influence of atutonomic neuropathy on mortality in insulin-dependent diabetes. Q..I. Med. 79(290), 495 502. [3] Sampson, M.J., Wilson, S.. Karagiannis, P., Edmonds, M. and Watkins P. (1990) Progression of diabetic autonomic neuropathy over at decadc in insulin-dependent diabetics. Q. J. Med. 178, 635 645. [4] Ewing, D.J., Campbell. l.W. and Clarke, B.F. (1980) The natural history of diabetic autonomic neuropathy. Q. J. Med. 193, 95 108.
15] Rathmaun, W.. Ziegler, D., Jahnkc, M., ltaastert, B. and Gries, V.A. (19931 Mortality in diabetic patients with cardiovascular autonomic neuropath~. Diabetic Med. 10, 820 824. [6] Ncsto. R.W., Phillips. R.T., Kett. S.G, et all. (1988) Angina aud exertional myocardial ischemia in diabetic and nondiabetic patients: assessment by exercise thallium scintigraphy. Ann. Intern. Med. 108, 170 175. [7] Margolis, J.R.. Kannell, W.S., Feinleig, M., Dawber, T.R. and McNamara, P.M. (1973) Clinical features of tmrecognized myocardial infarction silent and symptomatic. Eighteen year follow-up: Ttle Framingham Study. Am. J. Cardiol. 32. 1 7. [8] Chiariello, M., lndolfi, C., Cotecchia. M.R., Sifola, C.. Romano, M. and Condorelli, M. (1985) Asymptomatic transient ST changes during ambulatory ECG monitoring m diabetic patients. Am. Heart J. 110, 529 534. [9] Juhanti Airaksinen, K.E. and Juhani Koislinen, M. (1992) Association between silent coronary artery disease, diabetes and autonomic neuropathy. Diabetes Care 15(2), 288 292. [10] Wagener, D.K., Soaks, J.M., LaPorte, R.E. and MacGregor, J.M. (1982) The Pittsburgh study of insulin-dependent diabetes mellitus: risk for diabetes among relatives of IDDM. Diabetes 31, 136 144. [11] Orchard, T.J., Dormam J.S., Maser, R.E. et al. (1990) Factors associated with the avoidance of severe complicalions after 25 years of insulin-dependent diabetes mellitus: Pittsburgh Epidemiology of Diabetes Complications Study-1. Diabetes ('are 13, 741 747. [12] Orchard, T.J., Dorman, J.S., Maser. R.E. et al. (1990) The prevalence of complications in insulin-dependent diabetes mellitus by sex and duration: Pittsburgh Epidemiology of Diabetes Complications Study-ll. Diabetes 39. 1116 1124. [13] Sundkvist. G. (1981) Autonomic nervous function m asymptomatic diabetic patients with signs of peripheral neuropathy. Diabetes Care 4, 529 534. [14] "['he Diabetes Control and Complications Trial Research Group {1987) Manual of Operations Ibr the Diabetes Control and Complications Trial. Washington, DC: US Dept of Commerce.
T.J. Or
DiaDc:c,~ Research am/ C/iHi~al Prac:icc 34 Suppl. (1996) S I 6 5 S171
[15] Early Treatment Diabetic Retinopalhy Study Coordinating Center (1980) Manual of Operations. Baltimore, M D: University of Maryland School of Medicine, Diabetic Retinopathy Coordinating Center: Chapters 5 and 18. [16] Ellis, D. and Becket, D.J. (1974) Editorial reply. J. Pediatr. 104, 324 325. [I 7] Orchard, T.J. on behalf of the Community Cardiowtscular Surveillance Project investigators (1985) Validation of coronary heart disease mortality data: the Community Cardiovascular Surveillance Project Pilot Experience. CVD Epid. Newslett. 157, 46. [18] Allain, C., Pooll, LS.. Chan, C.S.(}., Richmond, W. and Fu, P.C. {1974) Enzymatic determination of total serum cholesteroh Clin. Chem. 20, 470 475. [19] Bucolo, G. and David, H.(1973)Quantitativedctermination of serum triglyceridcs by usc of cnzylnes. Clin. Chem. 19, 476 482. [20] Warnick, G.R. and Albcrs, J.J. (1978) tteparin Mn -~ quantitation of high density lipoprotein cholesterol: an ultrafiltralion procedure l'or lipcmic samples. Clin. Chem. 24, 9{){) 904. [21] National Institutes of Heahh. Department of ttealth, Education and Welfare (19751 Lipid Research Clinics Program. Washington, DC: Government Printing Office: 9-50. Publication No. Nltt 75-628. [22] Fricdewald, W.T., Levy, R.I. and Ercdrickson, D.S. (1972) Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin. Chem. 18, 499 502. [23] Borhani, N.O., Kass. f-.H., l,angford, H.G.. Payne, G.H., Remington, R.D. and Stamler. J. (1976) The hypertension detection and follow-up program. Prey. Med. 5. 207 215. [24] Lloyd, C.E., Becker. D.. Ellis. D. and Orchard. T.J.
[25]
[26]
[27]
[28]
[29]
[3(I]
[31]
SI7I
(1996) Incidence of complications in insulin-dependent diabetes mellitus: A survival analysis. Am. J. Epidemiol. 143. 431 441, Wilson, R, and Orchard. T.J. (1994) llypertension is a major risk factor for distal symmetrical polyneuropathy (DSP) in insulin dependent diabetes mellitus. American Diabetes Association, New Orleans, Diabetes 43(I), =4-42. Borch-Johnsen, K., Andersen, P.K. and Deckert, T. (1985) The effect of proteinuria on relative mortality in type I (insulin-dependent) diabetes mellitus. Diabetologia 28, 590 596. Donahue, R.P. and Orchard, T.J. (1992) Diabetes mellitus and macrovascular complications. An epidemiological perspective. Diabetes Care 15(9), 1141 1155. Maser, R.E.. Pfeilcr, M.A., Dorman, J.S.. KuNer, L.H., Becker, D.J. and Orchard, T.J. (1990) Diabetic autonomic neuropathy and cardiovascular risk. Pittsburgh Epidemiology of Diabetes Complications Study 111. Arch. lntern. Med. 150, 1218 1222. Ewing, D.J., Boland. O., Neilson, J.M.M., Cho, C.G. and Clarke. B.F. (1991) Autonomic neuropathy, QT interval lengthening and unexpected deaths in male diabetic patients. Diabelologia 34. 182 185. The Diabetes Control and Complications Trial Research Group (1993) The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N. Engh ,I. Med. 329, 977 986. Lewis, E.J., Hunsicker, L.G., Bain, R.P. and Rohde, R.D. (1993) The effect of angiotensin-con\,erting-enzyme inhibition on diabetic nephropathy. N. Engh J. Med. 329, 1456 1462.